Knowledge acquisition using evolutionary algorithms, Intelligent pattern recognition, EMG based pointing devices, Human sensing, Intelligent interface, signal processing (neural network, Evolutionary Systems, Knowledge Acqisition, Rule Generation, Image Understanding, Intelligent Systems, human sensing, digital signal processing, Statistical Learning Algorithms)
Book / Paper
Book:
1.
Stephen Githinji Karungaru, Kenji Terada and Minoru Fukumi : Character Recognition from Virtual Scenes and Vehicle License Plates using Genetic Algorithms and Neural Networks, InTech Inc, Oct. 2012.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Neural Networks and 3D Edge genetic Template Matching for Real Time Face Detection and Recognition, Chapter 9, Idea Group Inc, Oct. 2006.
3.
Minoru Fukumi : Handbook of fuzzy and soft computing, KYORITSU SHUPPAN.CO.,LTD, Tokyo, Jan. 2000.
Ulgen Figen, Norio Akamatsu and Minoru Fukumi : On-Lione Shape Recognition with Incremental Training Using a Neural Network with Binary Synaptic Weights, CRC Press, Jul. 1996.
Academic Paper (Judged Full Paper):
1.
Shafiq Muhammad Ibrahim, Rahayu Seri Kamat and Minoru Fukumi : Regression Analysis of Heart Rate for Driving Fatigue using Box-Behnken Design, Journal of Mechanical Engineering, Vol.21, No.1, 163-176, 2024.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Electroencephalogram Analysis Method to Detect Unspoken Answers to Questions Using Multistage Neural Networks, IEEE Access, Vol.11, 137151-137162, 2023.
(Summary)
Braincomputer interfaces (BCI) facilitate communication between the human brain and computational systems, additionally offering mechanisms for environmental control to enhance human life. The current study focused on the application of BCI for communication support, especially in detecting unspoken answers to questions. Utilizing a multistage neural network (MSNN) replete with convolutional and pooling layers, the proposed method comprises a threefold approach: electroencephalogram (EEG) measurements, EEG feature extraction, and answer classification. The EEG signals of the participants are captured as they mentally respond with yes or no to the posed questions. Feature extraction was achieved through an MSNN composed of three distinct convolutional neural network models. The first model discriminates between the EEG signals with and without discernible noise artifacts, whereas the subsequent two models are designated for feature extraction from EEG signals with or without such noise artifacts. Furthermore, a support vector machine is employed to classify the answers to the questions. The proposed method was validated via experiments using authentic EEG data. The mean and standard deviation values for sensitivity and precision of the proposed method were 99.6% and 0.2%, respectively. These findings demonstrate the viability of attaining high accuracy in a BCI by preliminarily segregating the EEG signals based on the presence or absence of artifact noise and underscore the stability of such classification. Thus, the proposed method manifests prospective advantages of separating EEG signals characterized by noise artifacts for enhanced BCI performance.
(Keyword)
Answer to question / convolutional neural networks / electroencephalogram / multistage neural networks, / personal model / support vector machine
Shafiq Muhammad Ibrahim, Kamat Rahayu Seri, Syamimi Shamsuddin and Minoru Fukumi : Mathematical Regression Analysis of Oxygen Saturation for Driving Fatigue using Box-Behnken Design, International Journal of Emerging Technology and Advanced Engineering, Vol.12, No.9, 23-29, 2022.
(Summary)
Malaysia have become an alarming concern. Mathematical modelling is the process of describing a real-world issue in mathematical terms to understand the original issue. Hence, this paper aims to develop a mathematical regression model to predict the relationship between five input variables namely (i) driving duration, (ii) driving speed, (iii) body mass index (BMI), (iv) types of roads and (v) gender and an output response (oxygen saturation level) as the causes of driving fatigue. The regression analysis utilized Box-Behnken design method by Design Expert (6.0.8) software. The results revealed that the Prob > F values for all input variables were less than 0.01%, implying that all the variables were significant in influencing the oxygen saturation level. The regression model was validated to determine its accuracy in predicting the output response. The analysis presented excellent prediction accuracy as the model was capable to predict the data within 95% predictive interval, which met the minimum quantitative condition of 90% predictive interval. Furthermore, the residual errors were less than 10%, indicating that the model has excellent accuracy in predicting the oxygen saturation. The model prediction is expected to be useful in guiding researchers and policy makers in road safety field to take measures in minimizing traffic accidents due to driving fatigue.
Ani Firdaus Mohammad, Kamat Rahayu Seri, Minoru Fukumi, Minhat Mohamad, Abdullah Abu and Husin Kalthom : Designing a graphical user interface for the decision support system of driving fatigue, International Journal of Advanced Mechatronic Systems, Vol.9, No.1, 30-37, 2021.
(Summary)
The paper presents the continuity study from the previous work, which designing the graphical user interface (GUI) for a decision support system (DSS) of driving fatigue. As driving fatigue has been recognized as one of the significant contributory factors to road accidents and fatalities in Malaysia, the author developed the DSS that providing analysis, and proving solutions and recommendations to the road users. In other words, the DSS acts as the advisory and decision-maker tool. In designing the GUI for a DSS, the Django based on Python programming language was used by the authors. There are five main GUI that has been designed in this study: admin GUI user profile and driving information GUI, regression model GUI, risk factor analysis GUI and superuser GUI. Further testing and validation of the graphical user interface for the DSS are suggested before it is used commercially.
Ani Firdaus Mohammad, Kamat Rahayu Seri, Minoru Fukumi and Noh Azila Nor : A Critical Review on Driver Fatigue Detection and Monitoring System, International Journal of Road Safety, Vol.1, No.2, 53-58, 2020.
(Summary)
This paper reviews existing and future fatigue detection and monitoring systems. Over the past few years, there has been an increase of interest in technologies, systems, and procedures to detect and monitor driver fatigue to reduce the number of road accidents. The driving activity has become more important as this medium is morepractical, faster,and cheaper in connecting humans around the world. However, driving activity can cause disastersor deathsto human in daily life as they get fatigued while driving. Driver fatigue is a vital contributor to road accidents. Studies show that 80.6% of road accidents are caused by human error which includes fatigue or drowsiness. Statistics indicate the need for a reliable driver fatigue detection and monitoring system, which could alert or warn the driver before anymishapshappens. Several approaches and methods have been developed to reduce the risk of fatigueamongdrivers, which uses the following measures:(1) vehicle-based measures; (2) behavioural measures; (3) physiological measures; (4) psychophysical measures;and(5) biomechanical measures. In this paper, the authorsbriefly review the literature on fatigue detection and monitoring systems. The findings from this review are discussed in the light of directions for future studies and the development of fatigue countermeasures.
Ani Firdaus Mohammad, Kamat Rahayu Seri and Minoru Fukumi : Development of Decision Support System via Ergonomics Approach for Driving Fatigue Detection, Journal of Social Sciences and Technical Education, Vol.1, No.1, 60-72, 2020.
(Summary)
Driving operation has become more important nowadays as this method becomes practical, faster, and cheaper to move humans from one location to another. However, driving activity can cause a human being to suffer tragedy or death in everyday life as they get exhausted while driving. Driver fatigue is a major contributing factor in road crashes. This paper's primary aim was to develop a decision support system (DSS) for the monitoring of driving fatigue. The decision support system seeks to provide systematic analysis and approaches to minimize the risk associated with driving exhaustion and the number of accidents involved. Four major stages involved as the cornerstone in the development of decision support system; knowledge acquisition, knowledge integration, development of driving fatigue strain index using fuzzy logic membership function, development of the fatigue driving decision support system (DSSfDF) model using the graphical user interface. The decision support system is an essential program for evaluating the risk factors which would significantly contribute to driving fatigue associated with driving activity. Furthermore, the decision support system offers users solutions and recommendations to minimize the number of road accidents in Malaysia.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Human-Wants Detection Based on Electroencephalogram Analysis During Exposure to Music, Journal of Robotics and Mechatronics, Vol.32, No.4, 724-730, 2020.
(Summary)
We propose a method to detect human wants by using an electroencephalogram (EEG) test and specifying brain activity sensing positions. EEG signals can be analyzed by using various techniques. Recently, convolutional neural networks (CNNs) have been employed to analyze EEG signals, and these analyses have produced excellent results. Therefore, this paper employs CNN to extract EEG features. Also, support vector machines (SVMs) have shown good results for EEG pattern classification. This paper employs SVMs to classify the human cognition into wants, not wants, and other feelings. In EEG measurements, the electrical activity of the brain is recorded using electrodes placed on the scalp. The sensing positions are related to the frontal cortex and/or temporal cortex activities although the mechanism to create wants is not clear. To specify the sensing positions and detect human wants, we conducted experiments using real EEG data. We confirmed that the mean and standard deviation values of the detection accuracy rate were 99.4% and 0.58%, respectively, when the target sensing positions were related to the frontal and temporal cortex activities. These results prove that both the frontal and temporal cortex activities are relevant for creating wants in the human brain, and that CNN and SVMare effective for the detection of human wants.
(Keyword)
wants detection / electroencephalogram / listening to music / convolutional neural network / support vector machine
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Individual Differences in Brain Activities When Human Wishes to Listen to Music Continuously Using Near-Infrared Spectroscopy, International Journal of Advanced Science and Technology, Vol.29, No.6, 807-813, 2020.
(Summary)
This paper introduces an individual difference in the activities of the prefrontal cortex when a person wants to listen to music using near-infrared spectroscopy. The individual differences are confirmed by visualizing the variation in oxygenated hemoglobin level. The sensing positions used to record the brain activities are around the prefrontal cortex. The existence of individual differences was verified by experiments. The experiment results show that active positions while feeling a wish to listen to music are different in each subject, and an oxygenated hemoglobin level is different in each subject compared to its value when a subject does not feel the wish to listen to music. The experiment results show that it is possible to detect a wish to listen to the music based on changes in the oxygenated hemoglobin level. Also, these results suggest that active positions are different in each subject because the sensitivities and how to feel on stimulus are different. Lastly, the results suggest that it is possible to express the individual differences as differences in active positions.
(Keyword)
Individual difference / near infrared spectroscopy / wish to listen to music / prefrontal cortex activity / music therapy
(Tokushima University Institutional Repository: 114822, Elsevier: Scopus)
9.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Japanese sign language classification based on gathered images and neural networks, International Journal of Advances in Intelligent Informatics, Vol.5, No.3, 243-255, 2019.
(Summary)
This paper proposes a method to classify words in Japanese Sign Language (JSL). This approach employs a combined gathered image generation technique and a neural network with convolutional and pooling layers (CNNs). The gathered image generation generates images based on mean images. Herein, the maximum difference value is between blocks of mean and JSL motions images. The gathered images comprise blocks that having the calculated maximum difference value. CNNs extract the features of the gathered images, while a support vector machine for multi-class classification, and a multilayer perceptron are employed to classify 20 JSL words. The experimental results had 94.1% for the mean recognition accuracy of the proposed method. These results suggest that the proposed method can obtain information to classify the sample words.
(Keyword)
Japanese sign language / Gathered image / mean image / convolutional neural network
Ani Firdaus Mohammad, Minoru Fukumi, Kamat Rahayu Seri, Minhat Bin Mohammad and Husain Kalthom : Development of driving fatigue strain index using fuzzy logic to analyze risk levels of driving activity, IEEJ Transactions on Electrical and Electronic Engineering, Vol.14, No.12, 1764-1771, 2019.
(Summary)
The objective of this study is to develop a driving fatigue strain index using fuzzy logic to analyze the risk levels of driving activity among road users. Driving fatigue is always related to the driving activity and has been identified as one of the vital contributors to the road accidents and fatalities in Malaysia. Therefore, the present article introduces the use of fuzzy logic for the development of strain index to provide the systematic analysis and propose an appropriate solution in minimizing the number of road accidents and fatalities. The development of strain index is based on the six risk factors associated with driving fatigue; muscle activity, heart rate, hand grip pressure force, seat pressure distribution, whole-body vibration, and driving duration. The data are collected for all the risk factors, and consequently the three conditions or risk levels are defined as safe, slightly unsafe, and unsafe. A membership function is defined for each fuzzy condition. IF-THEN rules were used to define the input and output variables, which correspond to physical measures. This index is a reliable advisory tool for providing analysis and solutions to driving fatigue problem, which constitutes the first effort toward the minimization of road accidents and fatalities.
Shun Yamamoto, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Verification of the Usefulness of Personal Authentication with Aerial Input Numerals Using Leap Motion, Advances in Science, Technology and Engineering Systems, Vol.4, No.5, 369-374, 2019.
(Summary)
With the progress of IoT, everything is going to be connected to the network. It will bring us a lot of benefits however some security risks will be occurred by connecting network. To avoid such problems, it is indispensable to strengthen security more than now. We focus on personal authentication as one of the security.As a security enhancement method, we proposed a method to carry out numeral identification and personal authentication using numerals written in the air with Leap motion sensor. In this paper, we also focus on proper handling of aerial input numerals to verify whether the numerals written in the air are helpful for authentication. We collect numerals 0 to 9 from five subjects, then apply three preprocessing to these data, learn and authenticate them by CNN (convolutional neural network) which is a method of machine learning. As a result of learning, an average authentication accuracy was 92.4%. This result suggests that numerals written in the air are possible to carry out personal authentication and it will be able to construct a better authentication system.
Shin-ichi Ito, Koyuki Orihashi, Momoyo Ito and Minoru Fukumi : A Gathered Images Analysis Method to Evaluate Sound Sleep, Journal of the Institute of Industrial Applications Engineers, Vol.7, No.1, 16-24, 2019.
(Summary)
This paper proposes a method to evaluate a sound sleep using an image gathering technique and itsanalysis techniques. The proposed method consists of three phases; gathered images generation, gathered images analysis and sound sleep evaluation. The gathered images designed to gather sleep postures and their changes are generated at 1 second, 10 seconds, 1 minute, 10 minutes, 1 hour intervals and all times, respectively. In the gathered image analysis, the gathered images are analyzed by calculating difference values among the gathered images of 10-minute and all times. Then, the sound sleep conditions are evaluated by visual inspection and analysis results. In order to show the effectiveness of the proposed method, we conduct experiments using real movies and their images. In experimental results, we confirm that there were sound sleep conditions, bad sleep conditions and borderline cases by checking subjective evaluation using questionnaire and generated gathered images visually. Moreover, we confirm that the calculated difference values among the gathered images of 10-minute and all times are different between sound sleep and other cases. Furthermore, the analyzed results show that the proposed method was successful in the sleep conditions classifications on four of five subjects. These results suggest that the gathered images analysis method is effective for evaluating whether sleep condition is sound sleep or not. In particular, it is important to calculate the difference values among the gathered images of 10-minute and all times to evaluate sleeping conditions.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : An Electroencephalogram Analysis Method to Detect Preference Patterns Using Gray Association Degrees and Support Vector Machines, Advances in Science, Technology and Engineering Systems, Vol.3, No.5, 105-108, 2018.
(Summary)
This paper introduces an electroencephalogram (EEG) analysis method to detect preferences for particular sounds. Our study aims to create novel braincomputer interfaces (BMIs) to control human mental (NBMICM), which are used to detect human mental conditions i.e., preferences, thinking, and consciousness, choose stimuli to control these mental conditions, and evaluate these choices. It is important to detect the preferences on stimuli. If the stimuli related to the preference can be detected, the NBCIMC can provide stimuli to the user based on their emotions by detecting their favorite stimuli. The proposed method adopted EEG recording technique, extraction techniques of EEG features and detection methods of preferences. EEG recording employs a simple electroencephalograph, for which the measurement position is the left frontal lobe (Fp1) of the brain. We assume that the differences of the EEG activities on the patterns of preference are expressed in the association between the changes of the power spectra on each frequency band of the EEG. To calculate the association, we employ the gray theory model. The EEG feature is extracted by calculating the gray association degree, then, the preferences are detect using a support vector machine (SVM). Experiments are conducted to test the effectiveness of this method, which is validated by a mean accuracy rate >88% on the favorite sound detection. These results suggest that the detection of subjects favorite sounds becomes easy when the EEG signals are analyzed while the gray associate degrees are used as the EEG feature and the SVM is used as the classifier.
(Keyword)
electroencephalogram / Preference / Favorite sounds / Simple electroencephalography / Gray association degree / Support vector machine
Taisei Watanabe, Tadahiro Oyama and Minoru Fukumi : 舌骨上筋群の筋電信号に基づくCNNを用いた舌動作と寡声母音の推定, IEEJ Transactions on Electronics, Information and Systems, Vol.138, No.7, 828-837, 2018.
(Summary)
In this paper, we propose a method to estimate the tongue motion direction and silent speech based on convolutional neural network (CNN) using the surface electromyogram (EMG) from the suprahyoid muscles. Conventional human machine interface (HMI) is difficult to use for users who are unable to freely move the muscles below the neck due to nerve damage or the like. Therefore, we have developed a method to estimate the tongue motion in 6 directions and 5 vowels of silent speeches from 4 channel EMG. As a result of verification experiment, we obtained averaged accuracy was about 81.2% in the estimation of the tongue directions and the silent speeches. Thus, it was suggested that simultaneous estimation is possible based on EMG measured from electrodes on the anterior neck region.
Ani Firdaus Mohammad, Kamat Rahayu Seri, Minoru Fukumi and Minhat Mohamad : Development of Driving Fatigue Strain Index for Reducing Accident Risk Among Drivers, International Journal of Electrical & Electronic Systems Research, Vol.12, 1-7, 2018.
(Summary)
Driving has become more important as this medium being practically, faster and cheaper in connecting human from one to other places. However, in some occurrences driving activity can cause disaster or death to human in daily life as they get fatigued while driving. Driving fatigue is one of the top contributor to the road crashes. Therefore this study is to develop a driving fatigue strain index (DFSI), collaborate with Decision Support System (DSS), to quantify the risk levels caused by driving activity, and to propose an appropriate solution in minimizing the number of road accidents caused by the driving fatigue. The decision support system provide fast and systematic analysis, and solutions to minimize the risk and the number of accidents associated with driving fatigue. The development of DFSI is based on risk factors associated with driving activity such as muscle activity, heart rate, hand grip force, seat pressure distribution, whole-body vibration, and driving duration. All risk factors are assigned with multipliers, and the DFSI is the output or result of those multipliers. The development of DFSI is essential to analyze the risk factors that would contribute significantly to discomfort and fatigue associated with driving. Besides, in the future this index will have a capability to recommend alternative solutions to minimize fatigue while driving.
(Keyword)
Driving Fatigue / Decision Support System / Road accident / driving fatigue strain index
Higasa Takashi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Specification Method of Character String Region in Augmented Reality, Journal of the Institute of Industrial Applications Engineers, Vol.6, No.2, 73-79, 2018.
(Summary)
This paper proposes a method to enter characters and/or character string in an augmented reality usinga gesture motion. The proposed method detects the region of character string using the gesture motion. It consistsof five phases; template generation, skin color detection, hand region detection, gesture motion extraction anddesignation of character string region. The template image consists of two fingers because a gesture is to takehold the tips of the first and second fingers. In the skin color detection, we extract the skin color on the basis ofvalues in saturation by using threshold processing. The hand region is detected by calculating areas and detectingthe area with the maximum value as a hand. The gesture motion is extracted using template matching. In orderto show the effectiveness of the proposed method, we conduct experiments for character string specification.
(Keyword)
Augmented Reality / HSV Color System / Gesture Motion / Quadrangle for character string region
Daiki Hiraoka, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Japanese Janken Recognition by Support Vector Machine Based on Electromyogram of Wrist, ECTI Transactions on Computer and Information Technology, Vol.11, No.2, 154-162, 2017.
(Summary)
In this paper, we propose a method which can discriminate hand motions. We measured an electromyogram of wrist by using 8 dry type sensors. We focused on four motions, such as ``Rock-Scissors-Paper'' and ``Neutral''. ``Neutral'' is a state that does not do anything. In the proposed method, we apply fast Fourier transformation (FFT) to measured EMG data, and then remove hum noise. Next, we combine values of sensors based on a gaussian function. In this gaussian function, variance and mean are 0.2 and 0, respectively. After that, we apply normalization by linear transformation to the values. Subsequently, we resize the values into range from -1 to 1. Finally, support vector machine (SVM) conducts learning and discrimination. We conducted experiments in three subjects. Discrimination accuracy of the proposed method for three subjects was 96.9%, 95.3%, 92.2%, respectively. It is clear that the gaussian function is robust to difference of sensor position because this function combines both adjacent channels of sensors.
(Tokushima University Institutional Repository: 113535)
18.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Consideration for Electroencephalogram Analysis using Self-Organizing Map Based on Learning Algorithm for Plural-Attribue Information, IEEJ Transactions on Electronics, Information and Systems, Vol.137, No.2, 302-309, 2017.
(Summary)
This paper discusses a method to detect electroencephalogram (EEG) patterns using a self-organizing map(SOM) based on a learning algorithm for plural-attribute information (SOMPA). The input data for SOMPA has two attributes which are EEG feature and individual feature. We set the EEG feature to main feature and individual feature to sub-attribute information. The winning node in the learning algorithm of SOMPA is determined by using main feature and sub-attribute information. In the preprocessing, we extract the EEG feature vector by calculating the time average on each frequency band which are , and , respectively. The individual feature is analyzed though the ego analysis using psychological testing. In order to prove the effectiveness of the proposed method, we conduct experiments using real EEG data. The experimental results show that the EEG pattern detection accuracy using SOMPA improves compared with the standard SOM.
Shin-ichi Ito, Momoyo Ito, Shoichiro Fujisawa and Minoru Fukumi : Method to Classify Matching Patterns between Music and Humans Mood Using EEG Analysis Technique Considering Personality, The Online Journal on Computer Science and Information Technology, OJCSIT, Vol.5, No.3, 341-345, 2015.
(Summary)
In this paper we introduce a method to classify matching patterns between music and human mood using an electroencephalogram (EEG) analysis technique and considering personality. We analyse the EEG of the left prefrontal cortex by single-point sensing. The EEG recording device uses dry-type sensors. The feature vector is created by connecting the personality quantification results and the EEG features. Egogramsthe Yatabe-Guilford personality inventory and a Kretschmer-type personality inventory are used to quantify personality. The EEG features are extracted using fast Fourier transform. Then, the matching patterns are classified using the k-nearest neighbour method. To show the effectiveness of the proposed method, we conduct experiments using real EEG data.
(Tokushima University Institutional Repository: 112404)
20.
Peng Zhang, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Development of Eye Mouse Using EOG signals and Learning Vector Quantization Method, Journal of the Institute of Industrial Applications Engineers, Vol.3, No.2, 52-58, 2015.
(Summary)
Recognition of eye motions has attracted more and more attention of researchers all over the world in recent years. Compared with other body movements, eye motion is responsive and needs a low consumption of physical strength. In particular, for patients with severe physical disabilities, eye motion is the last spontaneous motion for them to make a respond. In order to provide an efficient means of communication for patients such as ALS (amyotrophic lateral sclerosis) who cannot move even their muscles except eye, in this paper we proposed a system that uses EOG signals and Learning Vector Quantization algorithm to recognize eye motions. According to recognition results, we use API (application programming interface) to control cursor movements. This system would be used as a means of communication to help ALS patients.
Takako Ikuno, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Lost Property Detection by Template Matching using Genetic Algorithm and Random Search, Journal of the Institute of Industrial Applications Engineers, Vol.3, No.2, 59-64, 2015.
(Summary)
In this paper, we propose an object search method which is adapted to transformation of an object to be searched to detect lost property. Object search is divided into two types; global and local searches. We used a template matching using Genetic Algorithm (GA) in the global search. Moreover we use a random search in the local search. According to experimental results, this system can detect rough position of the object to be searched. The search accuracy obtained using the present method is 83.6%, and that of a comparative experiment using only GA is 42.1%. We have verified that our proposed method is effective for lost property detection. In the future, we need to increase search accuracy to search objects more stably. In particular, we need to improve local search.
Shin-ichi Ito, Momoyo Ito, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : Preference Analysis Method Applying Relationship between Electroencephalogram Activities and Egogram in Prefrontal Cortex Activities, --- How to collaborate between engineering techniques and psychology ---, International Journal of Advances in Psychology, Vol.3, No.3, 86-93, 2014.
(Summary)
This paper introduces a method of preference analysis based on electroencephalogram (EEG) analysis of prefrontal cortex activity. The proposed method applies the relationship between EEG activity and the Egogram. The EEG senses a single point and records readings by means of a dry-type sensor and a small number of electrodes. The EEG analysis adapts the feature mining and the clustering on EEG patterns using a self-organizing map (SOM). EEG activity of the prefrontal cortex displays individual difference. To take the individual difference into account, we construct a feature vector for input modality of the SOM. The input vector for the SOM consists of the extracted EEG feature vector and a human character vector, which is the human character quantified through the ego analysis using psychological testing. In preprocessing, we extract the EEG feature vector by calculating the time average on each frequency band: theta, low-beta, and high-beta. To prove the effectiveness of the proposed method, we perform experiments using real EEG data. These results show that the accuracy rate of the EEG pattern classification is higher than it was before the improvement of the input vector.
Shin-ichi Ito, Momoyo Ito, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : Preference Classification Method Using EEG Analysis Based on Gray Theory and Personality Analysis, The Online Journal on Computer Science and Information Technology, OJCSIT, Vol.4, No.3, 276-280, 2014.
(Summary)
This paper introduces a method to classify the preference patterns of sounds on the basis of an electroencephalogram (EEG) analysis and a personality analysis. We analyze the EEG of the left prefrontal cortex by single-point sensing. For EEG recording, a dry-type sensor and few electrodes were used. The proposed feature extraction method employs gray relational grade detection on the frequency bands of EEG and egogram. The gray relational grade is used for extracting the EEG feature. The egogram is extracted for quantifying the subject's personality. The preference patterns generated when the subject is hearing a sound are classified using the nearest neighbor method. To show the effectiveness of the proposed method, we conduct experiments using real EEG data. These results show that the accuracy rate of the preference classification using the proposed method is better than that using the method that does not to consider the subject's personality.
(Tokushima University Institutional Repository: 112402)
24.
Momoyo Ito, Kazuhito Sato and Minoru Fukumi : A Study of Safety Driving Support System focusing on Driver's Head Posture Categorization, International Journal of Engineering Research and Technology, Vol.2, No.9, 2702-2711, 2013.
(Summary)
In this paper, we analyze drivers head posture during safety verification and propose a method for classifying head posture using two types of unsupervised neural networks: Self-Organizing Maps (SOMs) and fuzzy Adaptive Resonance Theory (ART). The proposed method can generate the optimal number of cluster-generated labels for the target problem. We experimentally assess the effectiveness of the proposed method by adjusting the fuzzy ART network vigilance parameters. In addition, we indicate that drivers head posture during safety verification can be categorized according to their individual properties.
25.
Stephen Githinji Karungaru, Nakano Hitoshi and Minoru Fukumi : Road Traffic Signs Recognition using Genetic Algorithms and Neural Networks, International Journal of Machine Learning and Computing, Vol.3, No.3, 313-317, 2013.
Masato Miyoshi, Satoru Tsuge and Minoru Fukumi : Rhythm Features Based on Linear Predictive Coding of Energy Variations for Musical Mood Classification, Transactions of Information Processing Society of Japan, Vol.54, No.4, 1275-1287, 2013.
(Summary)
In this paper, we propose a novel rhythm feature, which we call Rhythm feature based on Linear Predictive Coding (RLPC), to improve mood classification performance. The proposed feature is extracted with Linear Predictive Coding (LPC) on energy variations of an audio signal and is able to represent periodicity of rhythm in musical audio signals. To evaluate the proposed feature in comparison with 5 conventional rhythm features, mood classification experiments were conducted for 7 moods. From these experimental results, average accuracy of the proposed feature was 83.7% and 1.1 point higher than that of the conventional features. In addition, in case of combining base features, which indicate intensity, timbre, and harmony features, with RLPC, average accuracy was 89.5%. The accuracy was 2.0 point higher than that of base features and 0.6 point higher than that of combining base features with the conventional features. From results of hypothesis test on each mood, accuracies of the proposed feature were significant against that of the 4 conventional features.
Yohei Takeuchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Nonlinear Learning Algorithm for Large Scale Datasets, International Journal of Engineering and Innovative Technology, Vol.2, No.7, 407-412, 2013.
(Summary)
Nonlinear feature generation techniques are among the most important processing tools in pattern recognition. As linear feature generation techniques, Principal Component Analysis (PCA) and Fisher Linear Discriminant Analysis (FLDA) have been developed and improved in various ways. Simple-FLDA (SFLDA) is an improved version of FLDA and has demonstrated good performance in recognition experiments. In SFLDA, eigenvectors spanning the eigenspace are obtained with simple iterative calculations, unlike the original FLDA which needs to solve eigenvalue problem of a covariance matrix created with all input samples. However, FLDA and SFLDA algorithms are both linear feature extraction methods. Therefore, the steps used for obtaining the eigenspace from an input space might not be suitable with respect to large, complex datasets. For adjustability to such datasets, the kernel trick is applied to many linear feature generation methods, turning them into nonlinear methods. In this paper, we propose Simple Kernel Discriminant Analysis (SKDA) for higher recognition performance by applying a kernel trick to the SFLDA algorithm. Similar to the SFLDA algorithm, SKDA algorithm is composed with simple calculations, but in recognition experiments using the UCI datasets as well as face image dataset; its features equal or surpass those of the SFLDA algorithm.
28.
Michihiro Jinnai, Neil Boucher, Minoru Fukumi and Taylor Hollis : A new optimization method of the geometric distance in an automatic recognition system for bied vocalisations, Acoustique & Techniques (TRIMESTRIEL D'INFORMATION DES PROFESSIONELS DE L'ACOUSTIQUE in French), No.Numero 68, 26-31, 2012.
(Summary)
We have been developing an automatic recognition system for bird vocalisations. Many biologists have been using the early 32 bit version of our system, and we have been working on a 64 bit version. The software segments a waveform of the bird vocalisation from a three-hour continuous recording and extracts the sound spectrum pattern from the waveform using the LPC spectrum analysis. Next, the software compares the sound spectrum pattern (the input pattern) with the standard pattern (that was extracted in advance) using a similarity scale. We use a new similarity scale called the ``Geometric Distance''. The Geometric Distance is more accurate than the conventional similarities in the noisy environment. In the 64 bit version, the software matches an input pattern with the 40,000 elements of the standard patterns per second and per processor, and it is 2.8 times faster than the conventional cosine similarity. In this paper, we introduce an automatic segmentation method of bird vocalisations and a new optimization method of the Geometric Distance. The new optimization method offers improvements of an order of magnitude over the conventional Geometric Distance.
29.
Yohei Takeuchi, Momoyo Ito and Minoru Fukumi : Novel Approximate Statistical Algorithm for Large Complex Datasets, International Journal of Machine Learning and Computing, Vol.2, No.5, 720-724, 2012.
(Summary)
In the field of pattern recognition, principal component analysis (PCA) is one of the most well-known feature extraction methods for reducing the dimensionality of high-dimensional datasets. Simple-PCA (SPCA), which is a faster version of PCA, performs effectively with iterative operated learning. However, SPCA might not be efficient when input data are distributed in a complex manner because it learns without using the class information in the dataset. Thus, SPCA cannot be said to be optimal from the perspective of feature extraction for classification. In this study, we propose a new learning algorithm that uses the class information in the dataset. Eigenvectors spanning the eigenspace of the dataset are produced by calculating the data variations within each class. We present our proposed algorithm and discuss the results of our experiments that used UCI datasets to compare SPCA and our proposed algorithm.
Seiki Yoshimori, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : Facial Impression Recognition Based on Facial Texture Information, Journal of Signal Processing, Vol.16, No.5, 419-426, 2012.
(Summary)
Facial consciousness and recognition have been studied for many years. However, these studies have only extracted the feature of facial impression. Some studies have treated the evaluation of facial impression. In those studies, liking, age, and gender were evaluated with respect to facial impression. Therefore, there are no details on whether we evaluate various general facial impressions in daily life. Moreover, the result of evaluation using those studied features depends on the facial-parts-extraction accuracy. Then, we use texture features for facial impression recognition in this study. The texture features can be extracted stably, because we need not specify their positions. We extract texture features in the mesh state. We can extract physical features by this technique without specifying the detailed location of features, unlike the conventional method. Finally, we show the effectiveness of our proposed method compared with previous studies.
(Keyword)
impression / face image / Gabor feature / random forest
Shin-ichi Ito, Yasue Mitsukura, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : Interindividual Difference Analysis in Prefrontal Cortex EEGs Based on the Relationship with Personality, Journal of Signal Processing, Vol.16, No.5, 443-450, 2012.
(Summary)
In this study, we propose a method to analyse interindividual differences in electroencephalograms (EEGs) taken while a subject listened to music, which is based on the relationship of these EEGs to human personality. The frequencies of the EEGs that are analysed have components that contain both significant and immaterial information as well as different levels of importance. We express the different levels of importance through the weight values of frequencies using a real-coded genetic algorithm. Then, the EEG patterns, which are determined based on the evaluation of the impression on the music, are detected using the $k$-nearest neighbour method. We assume that differences between detection results for EEG patterns with the highest and lowest recognition accuracies show interindividual differences. Moreover, ego analysis based on psychological testing is used to analyse personality, and the ego score is determined using a questionnaire. Finally, we discuss the relationship between personality and interindividual differences observed in experimental EEGs. An interesting tendency of a person with a combined ego type is that he or she has a unique response to negative stimuli compared with that of positive stimuli.
Stephen Githinji Karungaru, Kenji Terada and Minoru Fukumi : Detection of Abandoned Luggage and Owner Tracking at Sensitive Public Areas, IEEJ Transactions on Electronics, Information and Systems, Vol.133-C, No.1, 67-73, 2012.
(Summary)
Abandoned objects in sensitive congested public areas like airports or train stations pose a major security threat. Therefore, in this paper, to solve the problem, we propose a novel method for the detection of abandoned luggage, tracking its owner and extracting any other necessary information using multi-threshold pixel based dynamic background model and Earth Mover's Distance (EMD) signature matching. The public area selected for experiments is a train station. The background model is created by learning the color variance in all pixels by allowing multiple thresholds per pixel. After learning, pruning unnecessary thresholds improves the foreground extraction speed. Blob noise and outliers including shadows are deleted by a binarization method based on the Discriminant Analysis (DA) method. A color signature created using the HSV color space that is fast to process, is matched using the EMD metric to track blobs. Once an abandoned object candidate is found, a slower but more accurate SURF algorithm is used to extract feature points for further tracking. Stationary objects after this phase are considered to be abandoned luggage. To prove the effectiveness of the proposed method, experiments are conducted using the i-Lids dataset (2007 IEEE International Conference on Advanced Video and Signal based Surveillance (AVSS 2007)) achieving a frame-based average accuracy of about 93%.
Stephen Githinji Karungaru, Kenji Terada and Minoru Fukumi : Automatic Background Updating for Abandoned Object Detection at Train Stations, International Journal of Machine Learning and Computing, Vol.2, No.5, 609-613, 2012.
(Summary)
In video surveillance using overhead cameras, it is very important to capture and regularly updated the background to enable accurate extraction of persons or objects of interest. However, in scenes with a lot of movement and stationary objects, for example a train station, it is not easy to update the background without including such objects. In this work, we investigate several objects features and combine them to maintain a stable background image. The features include the speed of the object, texture, shape, associations between persons and objects, etc. The data used is a subset of the i-Lids dataset that was captured for analyzing video systems. It is captured in a train station using one overhead camera. Each video segment is about three minutes long. Index TermsActive background, abandoned objects, shape matching.
Stephen Githinji Karungaru, Ishitani Atsushi, Shiraishi Takuya and Minoru Fukumi : A Simple Interface for Mobile Robot Equipped with Single Camera using Motion Stereo Vision, International Journal of Machine Learning and Computing, Vol.2, No.5, 633-639, 2012.
Stephen Githinji Karungaru, Yoshida Toshihiro, Seo Toru, Minoru Fukumi and Kenji Terada : Monotonous Tasks and Alcohol Consumption effects on the Brain by EEG Analysis using Neural Networks, International Journal of Computational Intelligence and Applications, Vol.11, No.3, 1250015, 2012.
(Summary)
An analysis of the Electroencephalogram (EEG) signals while performing a monotonous task and drinking alcohol using principal component analysis (PCA), linear discriminant analysis (LDA) for feature extraction and Neural Networks (NNs) for classification is proposed. The EEG is captured while performing a monotonous task that can adversely affect the brain and possibly cause stress. Moreover, we investigate the effects of alcohol on the brain by capturing the data continuously after consumption of equal amounts of alcohol. We hope that our work will shed more light on the relationship between such actions and EEG, and investigate if there is any relation between the tasks and mental stress. EEG signals offers a rare look at brain activity, while, monotonous activities are well known to cause irritation which may contribute to mental stress. We apply PCA and LDA to characterize the change in each component, extract it and discriminate using a NN. After experiments, it was found that PCA and LDA are effective analysis methods in EEG signal analysis.
Koji Kashihara, Momoyo Ito and Minoru Fukumi : Automatic system to remove unpleasant images detected by pupil-size changes., International Journal of Computer Science Issues, Vol.9, No.1, 68-73, 2012.
37.
Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : Study on Model for Predicting the Intra-Individual Difference in Left Prefrontal Pole Electroencephalogram Variability and Its Evaluation, Electronics and Communications in Japan, Vol.94, No.5, 9-16, 2011.
(Summary)
This paper introduces a novel statistical model to estimate an intra-individual difference in left prefrontal cortex electroencephalogram (EEG) activities, and a method for evaluating the proposed model. It is known that an EEG contains the individual characteristics. However, extraction of these individual characteristics has not been reported. The analyzed frequency components of an EEG can be classified as the components that contain significant number of features and the ones that do not contain any. From the viewpoint of these feature differences, we propose the model for extracting features of the EEG. The model assumes a latent structure and employs factor analysis by considering the model error as personal error. We consider the first factor loading that is calculated by eigenvalue decomposition as the EEG feature. Furthermore, we use k-nearest neighbor (kNN) algorithm for evaluating the proposed model and the extracted EEG features. In general, the distance metric used is Euclidean distance. It is possible that the distance metric used depends on the characteristic of the extracted EEG feature and on the subject. Therefore, depending on the subject, we use one of the three distance metrics: Euclidean distance, cosine distance and correlation coefficient. Finally, in order to show the effectiveness of the proposed model, we perform an experiment using real EEG data.
koki Abiko, Hironobu Fukai, Yasue Mitsukura, Minoru Fukumi and Masahiro Tanaka : Face detection using RBF network in AIBO, IEEJ Transactions on Electronics, Information and Systems, Vol.130-C, No.11, 2031-2038, 2010.
(Summary)
We propose a face-tracking system for AIBO by using the skin color, in the first step of acquiring users characteristics. In this paper, we focus on the human-face, which has many kinds of characteristic parts in a human. We detect faces using the Neural Network (NN) for the purpose of estimating whether one pixel is skin color or not. However, the hierarchical NN may give this system false recognition for unknown color of background. In this paper, we propose high generality face recognition system for AIBO using the radial basis function (RBF) network. Also, in order to show the effectiveness of the proposed method, we perform computer simulations. First of all, we can see the skin color recognition results using the RBF network. In various light conditions, we have the relatively good results of the skin color recognition. Furthermore, we show output value distribution in color space. We can see the possibility of dealing with unknown color using the RBF network. Moreover, we have achieved the skin color recognition for AIBO in a real system. We will show that AIBO can track skin color in many kinds of light conditions.
(Keyword)
AIBO / RBF network / Face detection / RBF network / Face detection
Hironori Takimoto, Seiki Yoshimori, Yasue Mitsukura and Minoru Fukumi : Hand Posture Recognition Robust for Posture Changing in Complex Background, Journal of Signal Processing, Vol.14, No.6, 483-490, 2010.
(Summary)
In this paper, a method for hand posture recognition, which is robust to changing hand posture in an actual environment, is proposed. Conventionally, a data glove device and a 3D scanner have been used for feature extraction from hand shape. However, the performance of each approach is adversely affected by changing hand posture. Therefore, we propose a posture fluctuation model for efficient hand posture recognition based on the 3D hand shape and color features obtained from a stereo camera. A large dictionary for posture recognition is built from various inclined hand images, which were auto-created from one scanned hand image, based on the proposed method. In order to show the effectiveness of the proposed method, performance and processing times for posture recognition and compared with those of conventional methods. In addition, we perform an evaluation experiment using Japanese sign language.
(Keyword)
image processing / hand posture recognition / range image / complex background
H.K. Choge, Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : A Local DCT-II Feature Extraction Approach for personal Identification Based on Palmprint, IEEJ Transactions on Electronics, Information and Systems, Vol.130-C, No.9, 1657-1666, 2010.
(Summary)
Biometric applications based on the palmprint have recently attracted increased attention from various researchers. In this paper, a method is presented that differs from the commonly used global statistical and structural techniques by extracting and using local features instead. The middle palm area is extracted after preprocessing for rotation, position and illumination normalization. The segmented region of interest is then divided into blocks of either 8×8 or 16×16 pixels in size. The type-II Discrete Cosine Transform (DCT) is applied to transform the blocks into DCT space. A subset of coefficients that encode the low to medium frequency components is selected using the JPEG-style zigzag scanning method. Features from each block are subsequently concatenated into a compact feature vector and used in palmprint verification experiments with palmprints from the PolyU Palmprint Database. Results indicate that this approach achieves better results than many conventional transform-based methods, with an excellent recognition accuracy above 99% and an Equal Error Rate (EER) of less than 1.2% in palmprint verification.
Hironori Takimoto, Seiki Yoshimori, Yasue Mitsukura and Minoru Fukumi : Invisible calibration pattern based on human visual perception, IEEJ Transactions on Electronics, Information and Systems, Vol.130-C, No.8, 1440-1447, 2010.
(Summary)
In this paper, we propose an arrangement and detection method of an invisible calibration pattern based on characteristics of human visual perception. A calibration pattern is arranged around contents where invisible data is embedded, as some feature points between an original image and the scanned image for normalization of the scanned image. However, it is clear that conventional methods interfere with page layout and artwork of contents. Moreover, conventional visible patterns show a third person the position of embedded data. Therefore, visible calibration patterns are not suitable for security service. The most important part of human visual perception in the proposed method is the spectral luminous efficiency characteristic and the chromatic spatial frequency characteristic. In addition, a back ground color in surrounding of contents is not restricted to uniform color by using the proposed calibration pattern. It is suggest that the proposed method protect page layout and artwork.
Michihiro Jinnai, Satoru Tsuge, Shingo Kuroiwa and Minoru Fukumi : A New Geometric Distance Method to Remove Pseudo Difference in Shapes, International Journal of Advanced Intelligence (IJAI), Vol.2, No.1, 119-144, 2010.
(Summary)
In our previous paper, a new similarity scale called the Geometric Distance was proposed. With the conventional geometric distance algorithm, there are the following three shortcomings. 1. Since standard and input patterns are normalized to have the same area, a pseudo difference in shapes occurs between them. 2. Since ``shape variation'' is calculated in each combination of the standard and input patterns, the processing overhead increases when the number of standard patterns increases. 3. Since reference patterns are evaluated for each movement position of a normal distribution, the computational memory overhead increases when the number of components of standard and input patterns increases. To counter these shortcomings, a new geometric distance algorithm is proposed. 1. It is derived without normalization of the standard and input patterns, so that the pseudo difference in shapes is removed. 2. It reduces the processing overhead by separating the calculation of ``shape variation'' into registration process and recognition process. 3. It reduces the computational memory overhead by sharing a single reference pattern. Experiments in vowel recognition were carried out using the same voice data as the previous paper. At a mean of 5 dB SNR, the recognition accuracy improved from 78% to 82% over the conventional algorithm.
43.
Stephen Githinji Karungaru, Kamei Tomoaki, Fujiwara Masato, Norio Akamatsu and Minoru Fukumi : Vowel Recognition Using Akamatsu Integral and Differential Transforms, International Journal of Advanced Intelligence (IJAI), Vol.1, No.1, 125-140, 2009.
44.
Michihiro Jinnai, Satoru Tsuge, Shingo Kuroiwa, Fuji Ren and Minoru Fukumi : New Similarity Scale to Measure the Difference in Like Patterns with Noise, International Journal of Advanced Intelligence (IJAI), Vol.1, No.1, 59-88, 2009.
(Summary)
A new similarity scale called the Geometric Distance, that numerically evaluates the degree of likeness between two patterns is proposed. Traditionally, the similarity scales known as the Euclidean distance and cosine similarity have been widely used to measure likeness. Traditional methods do not perform well in the presence of noise or pattern distortions. In this paper, a new mathematical model for a similarity scale is proposed which overcomes these limitations of the earlier models, while improving the overall recognition accuracy. Experiments in speech vowel recognition were carried out under various SNR levels in a variety of noisy environments. In all cases a significant improvementin recognition accuracy is demonstrated, with the improvement most pronounced in the noisiest conditions. In fact, at a SNR of 5 dB in a subway, the recognition accuracyimproved from 65% to 75% and at 20 dB SNR from 98.4% to 99.6% over the MFCC method. Numerical modeling of simple patterns is used to demonstrate the principles behind the Geometric Distance.
Hillary Kipsang Choge, Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Development of a Block based Palmprint Recognition Methodology Using the Discrete Cosine Transform, Australian Journal of Intelligent Information Processing Systems, Vol.10, No.4, 22-31, 2009.
(Summary)
Hand-based biometric applications based on the palmprint have recently become a popular choice due to their high acceptance by the general public. In this paper, a method is presented which applies the type-II Discrete Cosine Transform (DCT) in a novel way by extracting selected coefficients from small overlapping blocks of scanned, medium-resolution palmprint images to form a feature vector for personal identification. The middle palm area is extracted after rotation, position and illumination normalization during the preprocessing stage before dividing it into 8×8 pixel blocks. The type-II DCT is then used to transform the blocks, and only a small set of the coefficients selected from each block after ranking based on magnitude only. The selected coefficients are concatenated to create a compact feature vector that represents each palmprint. Experiments using a subset of the PolyU Palmprint Database show that this block-based approach effectively discriminates between palmprints, with excellent palmprint recognition results and an Equal Error Rate (EER) of less than 2% in palmprint authentication when using a 64×64 pixel input image block.
46.
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Fast Approximate Incremental Learning Algorithm based on of Simple-FLDA, Journal of Signal Processing, Vol.13, No.6, 515-523, 2009.
(Summary)
This paper presents an Incremental Simple-FLDA. The Incremental Simple-FLDA is a fast incremental learning algorithm based on Simple-FLDA. This algorithm need not hold all training samples because it enables update of an eigenvector according to incremental samples. Moreover, this algorithm has an advantage that it can calculate the eigenvector at high-speed because matrix calculation is not needed. We performed computer simulations using UCI datasets, EMG data and face data to verify the effectiveness of the proposed algorithm. As a result, its availability was confirmed from the standpoint of recognition accuracy and convergence performance of the eigenvector compared with the other methods.
(Keyword)
linear discriminant analysis / incremental learning / dimensional reduction / biological signal / face recognition
Stephen Githinji Karungaru, Minoru Fukumi, Norio Akamatsu and Takuya Akashi : Detection and Recognition of Vehicle License Plates Using Template Matching, Genetic Algorithms and Neural Networks, International Journal of Innovative Computing, Information and Control, Vol.5, No.7, 1975-1985, 2009.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru and Minoru Fukumi : Interactive Interface with Evolutionary Eye Sensing and Physiological Knowledge, IEEJ Transactions on Electronics, Information and Systems, Vol.129-C, No.7, 1288-1295, 2009.
(Summary)
The purpose of this study is to develop an interactive interface using eye movement to operate a welfare apparatus, such as a feeding device for an orthopedically-impaired person. A part of the purpose is to eliminate special calibration and re-calibration during the operation. Originalities of this proposed system are as follows: eye sensing with evolutionary processing and interactive operation screen based on some physiological knowledge. The proposed system uses a non-contact type interface by the eye movement. An iris is tracked and eye movement is measured by the evolutionary eye sensing (EES) method. An operation screen is divided into 9 areas, which has visual stimulation using the physiological knowledge. A user can select these areas by eye movement and decide by eye fixation. The effectiveness of the proposed system is evaluated by comparison experiments with 20 subjects. The results indicate that the proposed system is easy to use for first-timer, who can become proficient in operation just after a few exercises.
(Keyword)
computer vision / Welfare Apparatus System / human interface / Evolutionary Video Processing
Choge Kipsang Hillary, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : A DFT-Based Method of Feature Extraction for Palmprint Recognition, IEEJ Transactions on Electronics, Information and Systems, Vol.129-C, No.7, 1296-1304, 2009.
(Summary)
Over the last quarter century, research in biometric systems has developed at a breathtaking pace and what started with the focus on the fingerprint has now expanded to include face, voice, iris, and behavioral characteristics such as gait. Palmprint is one of the most recent additions, and is currently the subject of great research interest due to its inherent uniqueness, stability, user-friendliness and ease of acquisition. This paper describes an effective and procedurally simple method of palmprint feature extraction specifically for palmprint recognition, although verification experiments are also conducted. This method takes advantage of the correspondences that exist between prominent palmprint features or objects in the spatial domain with those in the frequency or Fourier domain. Multi-dimensional feature vectors are formed by extracting a GA-optimized set of points from the 2-D Fourier spectrum of the palmprint images. The feature vectors are then used for palmprint recognition, before and after dimensionality reduction via the Karhunen-Loeve Transform (KLT). Experiments performed using palmprint images from the `PolyU Palmprint Database' indicate that using a compact set of DFT coefficients, combined with KLT and data preprocessing, produces a recognition accuracy of more than 98% and can provide a fast and effective technique for personal identification.
Stephen Githinji Karungaru, Takuya Akashi, Nakano Miyoko and Minoru Fukumi : Hour-glass Neural Network based Daily Money Flow Estimation for Automatic Teller Machines, IEEJ Transactions on Electronics, Information and Systems, Vol.129-C, No.7, 1325-1330, 2009.
(Summary)
Monetary transactions using Automated Teller Machines (ATMs) have become a normal part of our daily lives. At ATMs, one can withdraw, send or debit money and even update passbooks among other many possi- ble functions. ATMs are turning the banking sector into a ubiquitous service. However, while the advantages for the ATM users (Financial institution customers) are many, the Financial institution side faces an uphill task in management and maintaining the cash flow in the ATMs. On one hand, too much money in a rarely used ATM is wasteful, while on the other, insuficient amounts would adversely affect the customers and may result in a lost business opportunity for the financial institution. Therefore, in this paper, we propose a daily cash flow estimation system using neural networks that enables better daily forecasting of the money required at the ATMs. The neural network used in this work is a fiveve layered hour glass shaped structure that achieves fast learning, even for the time series data for which seasonality and trend feature extraction is difficult. Feature extraction is carried out using the Akamatsu Integral and Differential transforms. This work achieves an average estimation accuracy of 92.6%.
(Keyword)
ATMs / Neural Networks / Cash Flow Estimation / Time series / Akamatsu Transform
Stephen Githinji Karungaru, Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Image Morphing and Warping: Application to Speech Simulation using a Single Image, Journal of Advanced Computational Intelligence and Intelligent Informatics, Vol.13, No.4, 441-446, 2009.
Yasue Mitsukura, Koji Sakamoto, Hironobu Fukai, Seiki Yoshimori, Seiji Ito and Minoru Fukumi : Color feature extraction of the regions using the GA for the scenery image retreival, IEEJ Transactions on Electronics, Information and Systems, Vol.129-C, No.4, 710-719, 2009.
(Summary)
Recently, the keyword image retrieval is widely studied. By using these technologies, we can obtain the images with the corresponding keywords easily. In case of conventional image search systems, we search according to the file names basically. However, filenames which is named are frequently incorrect. To resolve this problem, we propose the automatic keyword addition method for scene images. In this paper, there are two important points. One of them is the image segmentation method using the maximum distance algorithm (MDA). The other is the automatic keyword addition using the color feature of regions. The other is the color feature extraction of regions. In the image segmentation method, we propose the automatic decision method of parameters in the MDA. For this purpose, we investigate the relation between the optimal parameters and features of regions. In the color feature extraction of regions, we propose the genetic algorithm(GA). Moreover, in order to show the effectiveness of the proposed method, we show the simulation examples. According to the results of the simulations, we achieve the keyword addition for scene images.
(Keyword)
content-based image retrieval / scene image / image segmentation / content-based image retrieval / scene image / image segmentation
Fukai Hironobu, Hironori Takimoto, Yasue Mitsukura, Tanaka Toshihisa and Minoru Fukumi : A Design of Apparent Age Estimation System by the Empirical Mode Decomposition, Journal of Circuits, Systems, and Computers, Vol.18, No.8, 1481-1492, 2009.
(Summary)
Recently, the automation of the age estimation technique is hoped for in various fields. Therefore, we propose an apparent-age estimation system using empirical mode decomposition (EMD). Conventional study reported that the time-frequency features are important for age estimation. However, these cannot necessarily extract the time-frequency feature in detail, because the classical technique that have a relationship of trade-off between the time resolution and the frequency resolution are used. On the other hand, the EMD is the novel time-frequency analysis technique that do not have the relationship of trade-off between the time resolution and the frequency resolution. The EMD gives a time-frequency analysis decomposing a signal into several intrinsic mode functions (IMFs). The IMF together with their Hilbert transforms are called the Hilbert Huang spectrum, which leads to instantaneous frequency and amplitude. We use these features effectively for extracting human's age perception. We estimate the age by a neural network that learns pairs of face image and the Hilbert Huang spectrum. Furthermore, we compress the data for neural network by using the simple principal component analysis (SPCA). In order to show the effectiveness of the proposed method, computer simulations are done by the actual human data.
Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : A model for extracting personal features of an electroencephalogram and its evaluation method, IEEJ Transactions on Electronics, Information and Systems, Vol.129, No.1, 17-24, 2009.
(Summary)
This paper introduces a model for extracting features of an electroencephalogram (EEG) and a method for evaluating the model. In general, it is known that an EEG contains personal features. However, extraction of these personal features has not been reported. The analyzed frequency components of an EEG can be classified as the components that contain significant number of features and the ones that do not contain any. From the viewpoint of these feature differences, we propose the model for extracting features of the EEG. The model assumes a latent structure and employs factor analysis by considering the model error as personal error. We consider the EEG feature as a first factor loading, which is calculated by eigenvalue decomposition. Furthermore, we use a k-nearest neighbor (kNN) algorithm for evaluating the proposed model and extracted EEG features. In general, the distance metric used is Euclidean distance. We believe that the distance metric used depends on the characteristic of the extracted EEG feature and on the subject. Therefore, depending on the subject, we use one of the three distance metrics: Euclidean distance, cosine distance, and correlation coefficient. Finally, in order to show the effectiveness of the proposed model, we perform a computer simulation using real EEG data.
(Keyword)
electroencephalogram / personal feature / genetic algorithm / distance function / k-nearest neighbor method / electroencephalogram / personal feature / genetic algorithm / distance function / -nearestneighbormethod
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Fast Incremental Algorithm of Simple Principal Component Analysis, IEEJ Transactions on Electronics, Information and Systems, Vol.129, No.1, 112-117, 2009.
(Summary)
This paper presents a new algorithm for incremental learning, which is named Incremental Simple-PCA. This algorithm adds an incremental learning function to the Simple-PCA that is an approximation algorithm of the principal component analysis where an eigenvector can be calculated by a simple repeated calculation. Using the proposed algorithm, it is possible to update the eigenvector faster by using incremental data. We carry out computer simulations on personal authentication that uses face images and wrist motion recognition that uses wrist EMG by incremental learning to verify the effectiveness of this algorithm. These results were compared with the results of Incremental PCA that introduced incremental learning function to the conventional PCA.
Shin-ichi Ito, Yasue Mitsukura, Hiroko Miyamura, Takafumi Saito and Minoru Fukumi : Extraction of EEG Characteristics While Listening to Music and Its Evaluation Based on a Latency Structure Model with Individual Characteristics, Electronics and Communications in Japan, Vol.92, No.1, 9-17, 2009.
(Summary)
EEG is characterized by the unique and individual characteristics. Little research has been done to take into account the individual characteristics when analyzing EEG signals. Often the EEG has frequency components which can describe most of the significant characteristics. Then there is the difference of importance between the analyzed frequency components of the EEG. We think that the importance difference shows the individual characteristics. In this paper, we propose a new EEG extraction method of characteristic vector by a latency structure model in individual characteristics (LSMIC). The LSMIC is the latency structure model, which has personal error as the individual characteristics, based on normal distribution. The real-coded genetic algorithms (RGA) are used for specifying the personal error that is unknown parameter. Moreover we propose an objective estimation method that plots the EEG characteristic vector on a visualization space. Finally, the performance of the proposed method is evaluated using a realistic simulation and applied to a real EEG data. The result of our experiment shows the effectiveness of the proposed method.
Stephen Githinji Karungaru, Fukuda Keiji, Minoru Fukumi and Norio Akamatsu : Fingerprints Images Enhancement Using a Concavity and Convexity method, Biomedical Soft Computing and Human Sciences, Vol.14, No.1, 39-46, 2008.
59.
Tomita Yohei, Shin-ichi Ito, Koda Naoko, Jianting Cao, Yasue Mitsukura and Minoru Fukumi : Objectively Psychological Evaluation Using the EEG, Journal of Signal Processing, Vol.12, No.6, 465-472, 2008.
(Summary)
The animal assisted therapy (AAT) has been known to have psychological and social effects on human being. However, scientific research is insufficient yet. Because it is difficult to have done in the medical clinic. Furthermore, the government cannot allow as a medical care because of nothing of clear proof. Therefore, in order to show a scientific basis of the AAT effects, we analyze the electroencephalogram (EEG) in the case of using the AAT. The EEG is generated by the brain activity, so that there is the correlation between the EEG and mentality. First of all, we define the healing by questionnaires. Secondly, we extract the correlation between the EEG features and mentality. Finally, we investigate the correlation between EEG features and the healing.
60.
Fukai Hironobu, Hironori Takimoto, Yasue Mitsukura, Tanaka Toshihisa and Minoru Fukumi : Apparent Age Feature Extraction by Empirical Mode Decomposition, Journal of Signal Processing, Vol.12, No.6, 457-464, 2008.
61.
Choge Kipsang Hillary, Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Enhancement of Palmprint Images Using an Optimized Hexagonal Multilayer Perceptron Neural Network, Journal of Signal Processing, Vol.12, No.6, 449-456, 2008.
62.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru and Minoru Fukumi : Evolutionary Video Processing for Lips Tracking, International Journal of Intelligent Computing in Medical Sciences and Image Processing, Vol.2, No.2, 111-125, 2008.
(Summary)
In this paper, real-time tracking and information acquisition of lip region of a person during speech in active video scenes are addressed. The lip is deformed by speech, and the size and orientation are changed by camera motion. The difficulty is mainly due to change of these appearances of the lip. We use template matching with a genetic algorithm to overcome these problems. A high speed and accurate tracking scheme using Evolutionary Video Processing is proposed. Usually, a genetic algorithm is unsuitable for a tracking accuracy of 91.6% and an average processing time of 26.0 milliseconds per frame are achieved.
Stephen Githinji Karungaru, Minoru Fukumi, Takuya Akashi and Norio Akamatsu : Genetic Algorithms based Adaptive Search Area Control for Real Time Multiple Face Detection using Neural Networks, WSEAS Transactions on Signal Processing, Vol.4, 97-109, 2008.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : An adaptive graininess suppression method for restoration of color degraded images, IEEJ Transactions on Electronics, Information and Systems, Vol.127, No.12, 2093-2100, 2007.
(Summary)
Previous studies of image restoration for noise images were based on a mask processing. These conventional noise removal methods based on the mask processing have an issue of defining degradation to accompany a spacial processing. In this paper, we propose a graininess suppression method based on edge shape. In this method, we detect edges from an noise image and perform graininess suppression for this image based on edge information. On the edge detection, we execute an image transformation for an image that enables us to extract edge by making principal component images. Moreover, we use the canny edge detection operator with can detect a weak edge that relates to a real edge, and do not detect a lie edge. In the suppression process, we use Wiener filter that can restore an noise image without making a complete edge map and the original signal map. We demonstrated the effectiveness of the present method for the noise added images and confirmed it by means of computer simulation.
Shin-ichi Ito, Yasue Mitsukura, 宮村 浩子, 斎藤 隆文, Minoru Fukumi and Cao Jianting : Detection Method of Music to Match Users Mood in Prefrontal Cortex EEG Activity, INFORMATION, Vol.10, No.6, 889-901, 2007.
(Summary)
In this article, we propose a method for detecting music to match a users mood in prefrontal cortex electroencephalogram (EEG) activity. The frequencies of the EEG analyzed are the components that contain significant and immaterial information. We focused on the significant frequency combinations. These frequency combinations are thought to express personal features of EEG activity. The proposed method calculates the percentage of the spectrum of these frequency combinations and evaluates whether the music matches the users mood through a simple threshold processing. A genetic algorithm (GA) is used to specify the frequency of personal features on the EEG.
66.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru and Minoru Fukumi : Downsized Evolutionary Video Processing for Lips Tracking and Data Acquisitiong, JACIII SCIS&ISIS2006, Vol.11, No.8, 1030-1042, 2007.
(Summary)
In this paper, high-speed lips tracking and data acquisition of a talking person in natural scenes are presented. Our approach is based on the Evolutionary Video Processing. This method has a trade-off between accuracy and a processing time. To solve this problem, in this paper, we proposed Evolutionary Video Processing with automatic SD-Control. In our simulations, the effectiveness of the proposed method is verified by a comparison experiment. The proposed method improves the performance, speed and accuracy, from 68.4% to 86.2%. Furthermore, it is evaluated that our proposed method can continue to chase the lips region even in such a case. It is demonstrated that the lips region detection and tracking at high speed and with high accuracy is possible, with acquisition of its numerical geometric change information.
67.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A robust gender and age estimation under varying facial pose, IEEJ Transactions on Electronics, Information and Systems, Vol.127, No.7, 1022-1029, 2007.
(Summary)
This paper presents a method for gender and age estimation which is robust for facial pose changing. We propose a feature point detection method which is the Adapted Retinal Sampling Method (ARSM), and a feature extraction method. A basic concept of the ARSM is to add knowledge about the facial structure into the Retinal Sampling Method. In this method, feature points are detected based on 7 points corresponding to facial organ from face image. The reason why we used 7 points to basis of feature point detection is that facial organ is conspicuous in facial region, and it is comparatively easy to extract. As features which is robust for facial pose changing, a skin texture, a hue and a gabor jet are used for the gender and age estimation. For classification of gender and estimation of seriate age, we use a multi-layered neural network. Moreover, we examine the left-right symmetric property of the face concerning gender and age estimation by the proposed method.
(Keyword)
gender and age estimation / facial image processing / neural network / gender and age estimation / facial image processing / neural network
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru and Minoru Fukumi : High Speed Genetic Lips Detection by Dynamic Search Domain Control, IEEJ Transactions on Electronics, Information and Systems, Vol.127, No.6, 854-866, 2007.
(Summary)
In this paper, high-speed size and orientation invariant lips detection of a talking person in an active scene using template matching and genetic algorithms is proposed. As part of the objectives, we also try to acquire numerical parameters to represent the lips. The information is very important for many applications, where high performance is required, such as audio-visual speech recognition, speaker identification systems, robot perception and personal mobile devices interfaces. The difficulty in lips detection is mainly due to deformations and geometric changes of the lips during speech and the active scene by free camera motion. In order to enhance the performance in speed and accuracy, initially, the performance is improved on a single still image, that is, the base of video processing. Our proposed system is based on template matching using genetic algorithms (GA). Only one template is prepared per experiment. The template is the closed mouth of a subject, because the application is for personal devices. In our previous study, the main problem was trade-off between search accuracy and search speed. To overcome this problem, we use two methods: scaling window and dynamic search domain control (SD-Control). We therefore focus on the population size of the GA, because it has a direct effect on search accuracy and speed. The effectiveness of the proposed system is demonstrated by performing computer simulations. We achieved a lips detection accuracy of 91.33% at an average processing time of 33.70 milliseconds per frame.
Shin-ichi Ito, Yasue Mitsukura, Takafumi Saito, Hiroko Miyamura and Minoru Fukumi : EEG Characteristic Extraction Method of Listening Music and Objective Estimation Method Based on Latency Structure Model in Individual Characteristics, IEEJ Transactions on Electronics, Information and Systems, Vol.127, No.6, 874-881, 2007.
(Summary)
EEG is characterized by the unique and individual characteristics. Little research has been done to take into account the individual characteristics when analyzing EEG signals. Often the EEG has frequency components which can describe most of the significant characteristics. Then there is the difference of importance between the analyzed frequency components of the EEG. We think that the importance difference shows the individual characteristics. In this paper, we propose a new EEG extraction method of characteristic vector by a latency structure model in individual characteristics (LSMIC). The LSMIC is the latency structure model, which has personal error as the individual characteristics, based on normal distribution. The real-coded genetic algorithms (RGA) are used for specifying the personal error that is unknown parameter. Moreover we propose an objective estimation method that plots the EEG characteristic vector on a visualization space. Finally, the performance of the proposed method is evaluated using a realistic simulation and applied to a real EEG data. The result of our experiment shows the effectiveness of the proposed method.
Tadahiro Oyama, Yuji Matsumura, Stephen Githinji Karungaru and Minoru Fukumi : Feature Generation Method by Geometrical Interpretation of Fisher Linear Discriminant Analysis, IEEJ Transactions on Electronics, Information and Systems, Vol.127, No.6, 831-836, 2007.
(Summary)
This paper presents a new algorithm for feature generation, which is derived based on geometrical interpretation of the fisher linear discriminant analysis (FLDA). This algorithm (Simple-FLDA) is an approximation algorithm that calculates eigenvectors sequentially by an easy iterative calculation by expressing the maximization of variance between classes and minimization of variance in each class without the use of matrix calculation. We carry out computer simulations about recognition of wrist motion patterns by EMG measured from wrist and personal authentications that use face images to verify the effectiveness of this technique. The result was compared with the result of principal component analysis (Simple-PCA).
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru and Minoru Fukumi : Using Genetic Algorithm for Eye Detection and Tracking in Video Sequence, Journal of Systemics, Cybernetics and Informatics, Vol.5, No.2, 72-78, 2007.
(Summary)
We propose a high-speed size and orientation invariant eye tracking method, which can acquire numerical parameters to represent the size and orientation of the eye. In this paper, we discuss that high tolerance in human head movement and real-time processing that are needed for many applications, such as eye gaze tracking. The generality of the method is also important. We use template matching with genetic algorithm, in order to overcome these problems. A high speed and accuracy tracking scheme using Evolutionary Video Processing for eye detection and tracking is proposed. Usually, a genetic algorithm is unsuitable for a realtime processing, however, we achieved real-time processing. The generality of this proposed method is provided by the artificial iris template used. In our simulations, an eye tracking accuracy is 97.9% and, an average processing time of 28 milliseconds per frame.
72.
Stephen Githinji Karungaru, Minoru Fukumi, Norio Akamatsu and Akashi Takuya : Automatic Human Faces Morphing Using Genetic Algorithms Based Control Points Selection, International Journal of Innovative Computing, Information and Control, Vol.3, No.2, 247-256, 2007.
Yuji Matsumura, Yasue Mitsukura and Minoru Fukumi : Hybrid EMG recognition system using linear discriminant analysis and principal component analysis, Transactions of the Institute of Systems, Control and Information Engineers, Vol.20, No.2, 51-59, 2007.
Stephen Githinji Karungaru, Minoru Fukumi, Norio Akamatsu and Takuya Akashi : Optimizing Feature Extraction for the Camera Mouse using Genetic Algorithms, WSEAS Transactions on Computers, Vol.5, No.11, 2722-2725, 2006.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Genetic Algorithms Based On-line Size and Rotation Invariant Face Detection, Journal of Signal Processing, Vol.9, No.6, 497-503, 2005.
77.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Face Recognition in Color Images Using Neural Networks and Genetic Algorithms, International Journal of Computational Intelligence and Applications, Vol.5, No.1, 55-67, 2005.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : EMG Signal Recognition System Using Feature Vectors by Genetic Function Identification, Journal of Signal Processing, Vol.9, No.3, 243-254, 2005.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Using the Thresholds Function Method to Detect License Plates, The Journal of the Institute of Image Information and Television Engineers, Vol.59, No.1, 115-122, 2005.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Morphing Human Faces: Automatic Control Points selection and Color Transition, Trans. on Engineering Computing and Technolog, Vol.1, No.1, 224-227, 2004.
81.
Miyoro Nakano, Fukimo Yasukata and Minoru Fukumi : Recognition of Smiling Faces Using Neural Networks and SPCA, International Journal of Computational Intelligence and Applications, Vol.4, No.2, 153-164, 2004.
(Summary)
In this paper, in order to achieve high-speed PCA, the simple principal component analysis (SPCA) is applied to compress the dimensionality of portions that constitute a face. A value of cos θ is calculated using an eigenvector by SPCA as well as a gray-scale image vector of each picture pattern. By using neural networks (NNs), the difference in the value of cos θ between the true and the false (plastic) smiles is clarified and the true smile is discriminated. Finally, in order to show the effectiveness of the proposed face classification method for true or false smiles, computer simulations are done with real images. Furthermore, an experiment using the self-organisation map (SOM) is also conducted as a comparison.
Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Proposal of the EEG Analysis Method Using the Individual Characteristic of the EEG, IEEJ Transactions on Electronics, Information and Systems, Vol.124-C, No.6, 1259-1266, 2004.
Yuji Matsumura, Minoru Fukumi, Norio Akamatsu and Kazuhiro Nakaura : Recognition of wrist EMG signal patterns using neural networks, Journal of Intelligent and Fuzzy Systems, Vol.15, No.3-4, 165-171, 2004.
(Summary)
we propose a recognition system based on wrist movements by focusing on ElectroMyoGram (EMG), using the body signals generated by voluntary movements of subject muscles, as the initial stage for construction of a total operation device. This paper tries to recognize EMG signals using neural networks (NNs). The electrodes under the dry state are attached to wrists and then EMG signals are measured. These EMG signals are classified using NNs into seven categories: neutral, up and down, right and left, inside twist, outside twist. The NN learns the FFT spectra of these signals in order to classify them. Moreover, we introduce a modular structure of the NN for improving the recognition accuracy. Computer simulations show that our approach is effective to classifying the EMG signals. This system is partly implemented on a DSP learning board.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A feature extraction method for personal authentication using SPCA and RGA, ヒューマンインタフェース学会論文誌, Vol.5, No.4, 499-506, 2003.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Neuro rainfall forecast with data mining by real-coded genetical processing, IEEJ Transactions on Electronics, Information and Systems, Vol.123-C, No.4, 817-822, 2003.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Skin Color Correction through Illuminant Estimation Using Neural Networks and Analytical Methods, Journal of Signal Processing, Vol.7, No.1, 69-78, 2003.
Miyoko Nakano, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Recognition of true smile based on simple-PCA, Journal of Signal Processing, Vol.6, No.6, 431-437, 2002.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Human Face Detection in Visual Scenes Using Neural Networks, IEEJ Transactions on Electronics, Information and Systems, Vol.122-C, No.6, 995-1000, 2002.
Minoru Fukumi, Yasue Mitsukura, Kazuhiro Nakaura and Norio Akamatsu : An Evolutionary Approach to Rule Generation from Trained Neural Pattern Recognition Systems, Journal of SYSTEMS RESEARCH and INFORMATION SYSTEMS, Vol.10, 71-88, 2001.
(Summary)
A method of extracting rules from neural pattern recognition systems formed using an evolutionary algorithm is presented. The evolutionary algorithm used here is a random optimization method (ROM). In particular, deterministic mutation (DM) is introduced in ROM. It is performed on the basis of the result of neural network strucutre learning. The DM procedure can evolve a candidate of a solution to increase a ROM fitness function in a deterministic manner. In the paper iris data are used as inputs. ROM are utilized to reduce the number of connection weights in the neural network. The network weights survived after the ROM training represent rules to perform pattern classification for the iris data. The rules are then extracted from the networks in which hidden units use signum and sigmoid functions to produce binary outputs. It enables us to extract simple logical functions from the network. Simulation results show this approach can generate a simple network structure and as a result simple rules for pattern classification.
91.
Minoru Fukumi and Yasue Mitsukura : Knowledge Incorporation and Rule Extraction in Neural Networks, Trans. of SSPATJ Japan, Vol.12, 13-18, 2001.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Fast Face Detection Method Using the GA-based Threshold Method considering lighting condition, Trans. of SSPATJ, Vol.12, 19-23, 2001.
Xiaoying Tai, Minoru Fukumi and Kenji Kita : 教師あり学習によるベクトル空間情報検索モデルの精度改善, IEEJ Transactions on Electronics, Information and Systems, Vol.121-C, No.10, 1647-1653, 2001.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A detection method of face region in color images by using the lip detection neural network and the skin distinction neural network, IEEJ Transactions on Electronics, Information and Systems, Vol.121-C, No.1, 112-117, 2001.
Minoru Fukumi, Yasue Mitsukura and Norio Akamatsu : A method for universal rule generation by genetic algorithm with virus infection, IEEJ Transactions on Electronics, Information and Systems, Vol.121-C, No.1, 212-217, 2001.
Minoru Fukumi, Sigeru Omatu, Fumiaki Takeda and Toshihisa Kosaka : Rotation-Invariant Neural Pattern Recognition System with Application to Coin Recognition, Transactions of the Society of Instrument and Control Engineers, Vol.E-1, No.1, 171-186, 2001.
97.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Fog occurrence forecast bynusing LVQ with a genetical preprocessing, IEEJ Transactions on Electronics, Information and Systems, Vol.120-C, No.12, 2055-2061, 2000.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Design and evaluation of neural networks for coin recognition by using GA and SA, Transactions of the Society of Instrument and Control Engineers, Vol.36, No.11, 1003-1008, 2000.
Minoru Fukumi and Norio Akamatsu : A method to extract rules from neural networks formed using evolutionary algorithms, IEEJ Transactions on Electronics, Information and Systems, Vol.120-C, No.4, 529-535, 2000.
Minoru Fukumi and Norio Akamatsu : Rule extraction from neural networks formed using random optimization method with deterministic mutation, Transactions of the Society of Instrument and Control Engineers, Vol.34, No.8, 1060-1065, 1998.
Minoru Fukumi, Toshiki Yoshino and Norio Akamatsu : Designing a Neural Network Using a Genetic Algorithm with Deterministic Mutation and Partial Fitness, Journal of Intelligent and Fuzzy Systems, Vol.6, No.1, 17-25, 1998.
(Summary)
In this paper a method of designing a neural pattern recognition system for a rotated coin recognition problem using a genetic algorithm (GA) with deterministic mutation (DM) and partial fitness (PF) is presented. In this method, chromosomes of individuals in the GA are divided into several parts and their PF functions are evaluated for GA operations. Furthermore, the DM, which is based on neural network learning, is introduced. The DM can evolve chromosomes of individuals to increase their fitness functions in a deterministic manner. In the pattern recognition system described in this paper, the Fourier transform is used as a preprocessor which produces rotation invariant features. These features are recognized by a multilayered neural network. The GA is utilized to reduce the number of signals, Fourier spectra, and input to the neural network. This approach using the GA is a type of feature selection problem. It is shown that the present method is better than conventional GAs with respect to convergence in learning, and results in the formation of a small neural network
Minoru Fukumi and Norio Akamatsu : Genetic algorithm with deterministic mutation based on neural network learning, The Transactions of the Institute of Electronics, Information and Communication Engineers D-II, Vol.J80-D-II, No.10, 2800-2807, 1997.
(Summary)
( )
103.
Minoru Fukumi, Sigeru Omatu and Yoshikazu Nishikawa : Rotation-Invariant Neural Pattern Recognition System Estimating a Rotation Angle, IEEE Transactions on Neural Networks, Vol.8, No.3, 568-581, 1997.
(Summary)
A rotation-invariant neural pattern recognition system, which can recognize a rotated pattern and estimate its rotation angle, is considered. It is well-known that humans sometimes recognize a rotated form by means of mental rotation. The occurrence of mental rotation can be explained in terms of the theory of information types. Therefore, we first examine the applicability of the theory to a rotation-invariant neural pattern recognition system. Next, we present a rotation-invariant neural network which can estimate a rotation angle. The neural network consists of a preprocessing network to detect the edge features of input patterns and a trainable multilayered network. Furthermore, a rotation-invariant neural pattern recognition system which includes the rotation-invariant neural network is proposed. This system is constructed on the basis of the above-mentioned theory. Finally, it is shown that, by means of computer simulations of a binary pattern and a coin recognition problem, the system is able to recognize rotated patterns and estimate their rotation angle
Minoru Fukumi, Sigeru Omatu and Yoshikazu Nishikawa : A method to design a neural network by the genetic algorithm with patial fitness, Transactions of the Institute of Systems, Control and Information Engineers, Vol.9, No.2, 74-81, 1996.
Minoru Fukumi, Sigeru Omatu and Yoshikazu Nishikawa : Rotated pattern recognition and rotation angle estimation by neural networks, IEEJ Transactions on Electronics, Information and Systems, Vol.155-C, No.10, 1199-1207, 1995.
Minoru Fukumi, Sigeru Omatu and Yoshikazu Nishikawa : A neural architecture for pattern recognition insensitive to translation, scale and line thickness, Transactions of the Society of Instrument and Control Engineers, Vol.30, No.11, 1360-1367, 1994.
Minoru Fukumi, Sigeru Omatu, Fumiaki Takeda and Toshihisa Kosaka : Rotation-invariant neural pattern recognition system with application to coin recognition, Journal of the Society of Instrument and Control Engineers, Vol.33, No.2, 151-165, 1994.
Minoru Fukumi and Sigeru Omatu : Designing an architecture of a neural network for coin recognition by a genetic algorithm, IEEJ Transactions on Industry Applications, Vol.113-D, No.12, 1403-1409, 1993.
Minoru Fukumi, Sigeru Omatu, Fumiaki Takeda and Toshihisa Kosaka : Rotation invariant neural network with an edge detection network, IEEJ Transactions on Electronics, Information and Systems, Vol.112-C, No.8, 457-464, 1992.
Minoru Fukumi, Sigeru Omatu, Fumiaki Takeda and Toshihisa Kosaka : Rotation-Invariant Neural Pattern Recognition System with Application to Coin Recognition, IEEE Transactions on Neural Networks, Vol.3, No.2, 272-279, 1992.
(Summary)
In pattern recognition, it is often necessary to deal with problems to classify a transformed pattern. A neural pattern recognition system which is insensitive to rotation of input pattern by various degrees is proposed. The system consists of a fixed invariance network with many slabs and a trainable multilayered network. The system was used in a rotation-invariant coin recognition problem to distinguish between a 500 yen coin and a 500 won coin. The results show that the approach works well for variable rotation pattern recogni.
Minoru Fukumi and Sigeru Omatu : Neural Pattern recognition System Invariant to Rotation of Input Pattern and Ite Application to Coin Recognition, Transactions of the Institute of Systems, Control and Information Engineers, Vol.5, No.1, 9-17, 1992.
Minoru Fukumi, Kayo Nishino and Shigeru Omatu : Nonlinear Adaptive Filter with a Neural Network, Transactions of the Institute of Systems, Control and Information Engineers, Vol.4, No.10, 429-431, 1991.
113.
Minoru Fukumi, Sigeru Omatu and Masaru Teranishi : A New Neuron Model "CONE" with Fast Convergence Rate and Its Application to Pattern Recognition, Systems and Computers in Japan, Vol.22, No.1, 91-98, 1991.
Minoru Fukumi and Sigeru Omatu : New IIR-Adaptive Algorithms Based on Orthogonalizaiton Method, Electronics and Communications in Japan, Part 3 : Fundamental Electronic Science, Vol.73, No.12, 37-45, 1990.
Minoru Fukumi and Sigeru Omatu : pattern Recognition System Insensitive to Translation and Rotation Using Neural Netwoeks with Sigmoid Functions, Transactions of the Institute of Systems, Control and Information Engineers, Vol.3, No.11, 381-388, 1990.
Minoru Fukumi, Sigeru Omatu and Masaru Teranishi : A New Neuron Model "CONE" with fast Convergence Rate and Its application to Pattern Recognition, The Transactions of the Institute of Electronics, Information and Communication Engineers D-II, Vol.J73-D-II, No.4, 648-653, 1990.
Minoru Fukumi, Sigeru Omatu and Hosokawa Naofumi : Pattern Recognition System Insensitive to Translation and Rotation by Neural Network, IEEJ Transactions on Electronics, Information and Systems, Vol.110-C, No.3, 148-155, 1990.
Minoru Fukumi and Sigeru Omatu : A new neuron Model "CONE" and its Learning Algorithm, IEEJ Transactions on Electronics, Information and Systems, Vol.110-C, No.3, 191-197, 1990.
Minoru Fukumi and Sigeru Omatu : New IIR-Adaptive Algorithms Based on Orthogonalization Method, The Transactions of the Institute of Electronics, Information and Communication Engineers A, Vol.J73-A, No.1, 44-50, 1990.
Minoru Fukumi : A Method of Design a Rotation Invariant Neural Pattern Recognition System by a Genetic Algorithm, Bulletin of Faculty of Engineering, The University of Tokushima, Vol.41, 87-94, 1996.
2.
Minoru Fukumi, Sigeru Omatu and Yoshikazu Nishikawa : Rotation Invariant Neural Network Model Capable of Estimating a Rotation Angle, Bulletin of Faculty of Engineering, The University of Tokushima, Vol.40, 63-70, 1995.
Academic Letter:
1.
Toshiki Yoshino, Minoru Fukumi and Norio Akamatsu : Designing a Neural Network by a Randpm Optimization Method with Deterministic Mutation, Transactions of the Institute of Systems, Control and Information Engineers, Vol.10, No.6, 338-340, 1997.
Minoru Fukumi and Sigeru Omatu : A New Back-Propagation Algorithm with Coupled Neuron, IEEE Transactions on Neural Networks, Vol.2, No.5, 535-538, 1991.
(Summary)
A novel neuron model and its learning algorithm are presented. They provide a novel approach for speeding up convergence in the learning of layered neural networks and for training networks of neurons with a nondifferentiable output function by using the gradient descent method. The neuron is called a saturating linear coupled neuron (sl-CONE). From simulation results, it is shown that the sl-CONE has a high convergence rate in learning compared with the conventional backpropagation algorithm
Toshiya Morisue and Minoru Fukumi : 3-D Eddy Current Calculation Using the Magnetic Vector Potential", IEEE Trans. on Magnetics, IEEE Transactions on Magnetics, Vol.MAG-24, No.1, 106-109, 1988.
(Summary)
Two methods for using the magnetic vector potential for 3-D eddy current calculation are treated. One method uses the magnetic vector potential that is continuous over the entire region and generally accompanies the electric scalar potential. It has the advantage that no cutting is necessary for the multiply-connected-region problem. The other method uses the magnetic vector potential that is discontinuous across the interface surface between different media. This magnetic vector potential can be arranged so that the electric scalar potential does not appear in the equations when the conductivity is constant. It has the disadvantage that cutting is necessary for the multiply-connected-region problem. New boundary value problem formulations are given for both methods, precisely defining the interface and boundary conditions.
5.
Toshiya Morisue and Minoru Fukumi : 3-D Magnetostatic Field Calculation Using the Magnetic Vector Potential and Boundary Integral Equation Method, IEEE Transactions on Magnetics, Vol.23, No.5, 3311-3313, 1987.
(Summary)
The magnetic vector potential and boundary integral equation method is an effective and promising method for 3-D magnetostatic field and eddy current calculations since it can handle the unbounded air region and requires less computation time and memory than FEM. In this paper the magnetic vector potential method is formulated as a boundary-value-problem, defining the interface and boundary conditions precisely. The computed results of the 3-D magnetostatic field in an iron slab surrounded with a current-carrying conductor are compared favorably with the experimental data when the sharp edges of the slab are rounded appropriately in the simulation model.
Momoyo Ito, Daiki Fujiwara, Shin-ichi Ito and Minoru Fukumi : Fundamental Study on the Influence of Driver Distraction Level on Face Orientation Change at Intersections, Proceedings of 7th International Symposium on Future Active Safety Technology toward zero traffic accidents, Thu-PM1-B-5, Kanazawa, Nov. 2023.
2.
Hideyuki Mimura, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Personal Authentication and Recognition of Aerial Input Hiragana Using Deep Neural Network, Proceedings Volume 11794, Fifteenth International Conference on Quality Control by Artificial Vision, 1-7, Tokushima (Running Virtually), May 2021.
(Summary)
We use Leap Motion and a deep neural network to perform personal authentication and character recognition of all hiragana characters entered in the air. We use Leap Motion to detect the index finger and store its trajectory as time series data. The input data was pre-processed to unify the data length by linear interpolation. For identification, the accuracy of Long Short Term Memory (LSTM) was compared with Support Vector Machine (SVM). As a result, SVM and LSTM achieved 97.25% and 98.18% F-measure in character recognition, respectively. In personal authentication, SVM has an accuracy of 92.45%, False Acceptance Rate (FAR) was 0.73%, and False Rejection Rate (FRR) was 41.59%. On the other hand, LSTM had an accuracy of 96.13%, FAR of 1.73% and FRR of 14.55%. Overall, the LSTM performed better than the SVM.
(Keyword)
Biometrics / Personal Authentication / Leap motion / Aerial Input Hiragana / deep learning
Eisuke Yamamoto, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Separation of Compound Actions with Wrist and Finger Based on EMG, Proc. SPIE 11794, Fifteenth International Conference on Quality Control by Artificial Vision, 1-7, Tokushima (Running Virtually), May 2021.
(Summary)
In this paper, we propose to measure the EMGs of the wrist and fingers using dry-type sensors worn near the wrist, and to separate the measured data into wrist and finger EMGs by using independent component analysis (ICA). Then we can confirm the EMGs of the wrist and fingers from the complex motion and realize individual identification in more complex motions. The final goal of this study is to identify individual motions from complex motions. In this paper, as a preliminary step, the ICA is used to isolate compound motions and the validity of the method is evaluated. We measured the EMGs for three days and four motions. The results of the combination of FastICA, Infomax and JADE, respectively, were evaluated by the correlation coefficient with the original signal. The most accurate combination was FastICA + Infomax with an accuracy of 70.5%.
Tsubasa Fukui, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Motion Identification of fingerspelling by Wrist EMG Analysis, Proc. 2020 IEEE SYMPOSIUM SERIES ON COMPUTATIONAL INTELLIGENCE, 1739-1744, Running Virtually, Dec. 2020.
(Summary)
Recent years, interfaces using biometric information are progressing. Electromyogram(EMG) has been used in a variety of situations. Many studies have measured EMG in the shoulders and arms where there is a lot of muscle mass In addition, wet type sensors have been often used. However those are inconvenient to use in everyday life and high cost. In this research we measure wrist EMG for convenience and cost. Currently researches have been done on the wrist EMG motion identification and personal identification. These studies have conducted simple movements and a large number of electrodes for discrimination. Furthermore authentication by password sequence with gestures has not been done. In this paper we propose to realize motion identification and personal authentication with complex movements using a small number of electrodes. The measured data was preprocessed such as removing noise and smoothing. We compared the accuracies obtained using Support Vector Machine(SVM) and Long Short term memory(LSTM) for motion identification and authentication. The accuracies obtained using SVM and LSTM were 60.4%and 62.4%, respectively. In this case the number of data was small. It is therefore necessary for increasing the number of data to perform deep learning.
Kazuki Nagatomo, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Finger Motion Identification Based on Wrist EMG Analysis Using Machine Learning, Proc. of International Conference on System Science and Engineering 2020, 522-523, Sep. 2020.
(Summary)
In this paper we identify 3 pattern motions that are Rock-scissors-paper finger motions. Conventional work recognized such motions using EMG in a stable state. We try to identify the motions by using the short-time wrist EMG at the beginning of movement and SVM. As a result, we obtained identification accuracy of 82%.
(Keyword)
EMG / Finger motion / machine learning / SVM
6.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Japanese Sign Language Classification Using Gathered Images and Convolutional Neural Networks, Proceedings of 2020 IEEE 2nd Global Conference on Life Sciences and Technologies (LifeTech), 349-350, Kyoto, Mar. 2020.
(Summary)
A method to classify Japanese sign language (JSL) words using a gathered image generation method and a convolutional neural network (CNN) is proposed. The JSL words consist of words that are often used in information queries. Gathered images are generated based on the difference between the first image, which indicates the start position of a JSL word and target images, which indicate the motion position for a JSL word. The CNN is used to extract features from the gathered images. The JSL words are classified using a support vector machine. To show the effectiveness of the proposed method, we conducted experiments and computer simulations. We confirmed that the mean recognition accuracy for 10, 20, and 42 JSL words was 99.2%, 94.3, and 86.2%, respectively.
(Keyword)
Japanese sign language / gathered image / convolutional neural networks / communication tool / welfare system
CHUNYU GUO, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Personal Authentication by Walking Motion using Kinect, Proc. of ISPACS 2019, 1-2, Taipei, Dec. 2019.
(Summary)
In recent years, with the rapid development of the information society, the importance of personal authentication has become higher and higher. This paper focuses on the use of a Kinect sensor to obtain walking characteristics for personal authentication. In terms of the proposal method, Kinect is used to obtain body's physical feature quantity, such as the angle of joint bending when a person walks, the displacement of coordinates. In terms of learning recognition, the support vector machine and the obtained feature amount are used for personal authentication. We measured 3 subjects data 5 times a day for 4 days, and obtained an average recognition accuracy of 77.4% using crossvalidation.
(Keyword)
Personal authentication / SVM / Walking motion / Kinect
Shan Xiao, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Study on Discrimination of Finger Motions based on EMG signals, Proc. of ISPACS 2019, 1-2, Taipei, Dec. 2019.
(Summary)
In recent years, biological signals have attracted attention as tools for human interfaces. Researches on biological signals have been actively conducted. In this paper, we propose a method which distinguishes ten motions, such as One Two Three Four Five Six Seven Eight Nine and Ten by measured the electromyogram of the wrist. We measure data by installing 8 dry type sensors on the right wrist. We carry out frequency analysis using FFT and try to take 3 kinds of methods to remove noise. Finally, we use Support Vector Machine (SVM) for identification and classification. We conducted experiments with four subjects. In the experimental result, the accuracy of finger motions recognition was 65%. In the future, we will also add more methods to remove noise, and try to find other methods to improve the accuracy in the research.
Kazuki Shimamoto, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Nail Image Analysis Method to Evaluate Accumulated Stress Using Fuzzy Reasoning, Proc. of ISPACS 2019, 1-2, Taipei, Dec. 2019.
(Summary)
In this paper, we propose a nail image analysis methodtoevaluateaccumulatedstressusingfuzzyreasoning.The proposed method consists of three stages: measurement, feature extraction, and stress evaluation. In the measurement, we take a nail image. In the feature detection, we extract the lunula of the nail to calculate its height. In the stress evaluation, we evaluate accumulated stress using fuzzy reasoning. The experimental results suggest that the proposed method can determine the presence or absence of accumulated stress.
Yurika Fujii, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Detection of Dangerous Objects By Pan-tilt Camera, Proc. of ISPACS 2019, 1-2, Taipei, Dec. 2019.
(Summary)
Security cameras have increased in public facilities. The number of crimes has decreased by security cameras, but we will have too much data of cameras. In this paper, we have aim of security improvement. First, we search whether there are humans in images by OpenPose. We then obtain position of humans hands. Finally, we detect dangerous objects around the hands by image classification.
(Keyword)
Pan-tilt Camera / Transfer learning / machine learning / Deep learning / dangerous object
Misato Matsushita, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Verication of Regression Analysis of Muscle Fatigue Using Wrist EMG, Proc. of ISPACS 2019, 1-2, Taipei, Dec. 2019.
(Summary)
Muscles can cause injury by training to improve physical performance. However, there are few ways to assess muscle fatigue currently. Therefore, in this paper, muscle fatigue is evaluated using surface EMG(ElectroMyoGram). For discrimination, we used linear regression analysis and support vector regression, and performed comparative verication.
Shin-ichi Ito, Momoyo Ito, Shoichiro Fujisawa and Minoru Fukumi : Electroencephalogram Data for Classifying Answers to Questions with Neural Networks and Support Vector MachineNetworks, Proceedings of International Conference on Electronics and Signal Processing, ICESP2019, Hong Kong, Aug. 2019.
(Summary)
This paper proposes a method for classifying answers to conversational questions from electroencephalogram (EEG) data. The proposed method includes steps for EEG recording, feature extraction, and answer classification. For EEG measurements, this paper employs a simple electroencephalograph. The EEG signals from the frontal lobe are recorded. The EEG features are calculated by normalizing the EEG signals and using convolutional neural networks (CNN) for extraction. The answers to questions are then classified from the EEG features using a support vector machine. To show the effectiveness of the proposed method, we conducted experiments using real EEG data. The experimental results confirm that the mean recognition accuracy was 99% or more if the CNN features are individual to the subject. These results suggest that the answers to yes/no questions can be classified using EEG signals and that the EEG analysis technique using CNN and the support vector machine is suitable for extracting and classifying EEG features.
(Keyword)
electroencephalogram / answers of questions / convolutional neural networks / personal differences / human support system / human communication
13.
Kohei Sasada, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Proposal of Japanese Sign Language Motion Recognition Method using Leap Motion, Proc. of ICEAS 2019, 74-84, Honolulu, Aug. 2019.
(Summary)
In Japan, in recent years, many people suffer from hearing and language related disorders. Sign language is then one of their main means of communication. As their opinion, there is a demand for smooth communication means. In recent years, researches using motion sensors are in progress. In particular, we focus on Leap Motion as a motion sensor in this research. The final goal is to build a sign language recognition system, and this paper proposes a sign language recognition method as the first step. The measured data using the Leap Motion sensor are subjected to preprocessing such as vectorization, distance transformation, noise processing, data length change, and correction of operation start position. An identification unit uses SVM and CNN to compare their accuracy. The result was 98.86% for SVM and 97.11% for CNN. SVM was 1.75% higher than CNN in accuracy. However, the CNN used in this experiment has a simple layer configuration, and its accuracy can be expected to be improved by changing the layer configuration. In addition, it is considered that the accuracy is improved by adding a feature amount or devising pre-processing as a measure for improving misidentification. In the future, we would like to introduce those methods and cope with sign languages other than Japanese sign language.
(Keyword)
Sign Language / Leap Motion / machine learning / Deep learning / CNN
14.
Kenta Matsumura, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Driver State Estimation Based on Visual and Heart Rate Statistical Features, Proc. of ICEAS 2019, 62-73, Honolulu, Aug. 2019.
(Summary)
Inattentive driving is one of the main causes of traffic accidents. It is required to develop a system for detecting this inattentive state from in-vehicle information and biological information. In this paper, we obtain visual information and heart rate information and describe the driver's state estimation. We verify the usefulness of the proposed method in a simulated environment with a driving simulator. Moreover, in order to aim at the installation of a real vehicle environment, we build a general model which doesn't choose the target person, and an individual model focusing on driving characteristics such as individual habits and investigate the usefulness of these models.
Ryosuke Saitoh, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Construction of Japanese Vowel Identification System Using Lips EMG, Proc. of ICEAS 2019, 41-49, Honolulu, Aug. 2019.
(Summary)
Recent years, biological signals have attracted much attention as a tool of human interface. Electromyogram(EMG) has been used in a variety of situations in particular. On the one hand, people who lost their voices due to vocal disorder communicate by alternative means of voice. In recent years, researches for detecting speech by electromyography analysis and image analysis have been actively conducted. Therefore, in this paper, we measure EMG by attaching dry type sensors to facial muscles, and identify Japanese vowels. A method proposed in this paper consists of an input, a preprocessing, and a learning identifying sections. We attach dry type sensors to muscles around the lips and measure EMG signals. We use a convolutional neural network(CNN) for learning and identification. In addition, we try to use a support vector machine(SVM) for comparison. The average identification accuracy by CNN was 67.4%. On the other hand, the average identification accuracy by SVM was 70.2%. In future work, we will try to increase the number of data and improve CNN accuracy. Therefore, it is necessary to ameliorate a layer configuration in CNN.
(Keyword)
Deep Learning / Lips EMG / Japanese Vowel Identification
16.
Shun Yamamoto, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : An authentication system for aerial input numerals using Leap motion and CNN, Proc. of ICEAS 2019, 29-40, Honolulu, Aug. 2019.
(Summary)
As information technology has advanced in recent years, services which include personal authentication systems such as ATM are increasing. Current main personal authentication systems include IC cards, passwords, and biometrics authentication such as fingerprint authentication. However, there are several problems in these systems. Therefore, better systems are needed. As such systems, we propose a method to write numerals in the air using the Leap motion and to carry out personal authentication from such aerial handwriting data. We try to authenticate numerals 0 to 9 which are written by three subjects. After applying some pre-processing to inputs, learning and identification are carried out using CNN which is a method of machine learning. As a result, average identification accuracy was 92.7%. From this result, it is suggested that input numerals in the air can be authenticated and there is a possibility to construct a new personal authentication system.
(Keyword)
Deep Learning / Leap Motion / Authentication
17.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method of Classifying Japanese Sign Language using Gathered Image Generation and Convolutional Neural Networks, Proceedings of International Conference on Pervasive Intelligence and Computing, PICom2019, 868-871, Fukuoka, Aug. 2019.
(Summary)
This paper proposes a method for classifying Japanese sign language (JSL) using a combined gathered image generation technique and a convolutional neural network (CNN) approach. In the combined gathered image generation, the maximum difference from the previous and next images is calculated for each block, and the block information that had maximum difference was embedded into an image on all blocks. After information on all images has been gathered into single words, the CNNs are used to extract features for the classification of JSL words. A multi-class support vector machine (SVM) is then used to classify words related to greeting and requesting. The mean and the standard deviation of the recognition accuracy of the proposed method were experimentally shown to be 84.2% and 4%, respectively. These results suggest that it is possible to obtain information for classifying 10 JSL words using the proposed combined gathered image generation and CNN approach.
Yurika Fujii, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Detection of Dangerous Objects By Pan-tilt Camera, Proceedings of The Fifth International Conference on Electronics and Software Science ICESS2019, Japan, 2019, 61-70, Takamatsu, Aug. 2019.
(Summary)
Security cameras are increasing in public facilities. The main reason is improvement of security. Furthermore, security cameras help quick arrest of criminals. It is, therefore, important that we install security cameras. However, we expect to obtain too much data of movie by increasing the number of cameras too. Security guard cannot watch all movie of all cameras always. For these reasons, we thought not only getting data but also security improvement. If humans have dangerous objects, the movies situation should define them as dangerous. First, we detect humans in camera's images. Second, if we detect dangerous objects in human detection area, we think that there are dangerous humans. We use SSD to detect humans and dangerous objects.
(Keyword)
SSD / security camera / image processing / Deep learning
19.
Misato Matsushita, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Multiclass Classification and Regression Analysis of Muscle Fatigue Using Wrist EMG, Proceedings of The Fifth International Conference on Electronics and Software Science ICESS2019, Japan, 2019, 90-95, Takamatsu, Aug. 2019.
(Summary)
Muscles can cause injury by training to improve physical performance. However, there are few ways to assess muscle fatigue currently. Therefore, in this paper, muscle fatigue is evaluated using surface EMG(ElectroMyoGram). The proposed method in this research consists of 4 parts: Measurement, Pre-processing, Feature extraction, and Learning identification parts. The effectiveness of the proposed method is demonstrated in two ways, classification and regression analysis, and comparative verification is conducted.
Kazuki Shimamoto, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Evaluate Accumulated Stress Using Nail Image, Proceedings of The Fifth International Conference on Electronics and Software Science ICESS2019, Japan, 2019, 12-17, Takamatsu, Aug. 2019.
(Summary)
In this paper, we propose a method to evaluate accumulated stress by extraction the height of the lunula of the nail from a nail image. The proposed method consists of three stages: measurement, preprocessing, and stress evaluation. In the measurement, we take a nail image. In the preprocessing, we extract the height of the lunula of the nail. Then, we carry out edge detection using a hue histogram in a rectangle. In the stress evaluation, we evaluate accumulated stress at 0 to 1 using fuzzy reasoning. In order to show the effectiveness of the proposed method, we conducted experiments. These results suggested that the difference between the minimum and maximum values of the height of the lunula while the experiment might be able to determine the presence or absence of accumulated stress.
(Keyword)
image processing / fuzzy reasoning / stress evaluation / nail image / social system
Kazuki Shimamoto, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Extract Change of Lunula of the Nail, Proc. of SAMCON 2019, TT9-2-1-TT9-2-4, Chiba, Mar. 2019.
(Summary)
This paper proposes a method to extract change of lunula of the nail. The proposed method consists of three phases; HSV color system conversion, finger detection, and rectangle search. In the HSV color system conversion, we converts the RGB color system into the HSV color system. Then, color components are divided into each component of Hue, Saturation, Value. In the finger detection, we use the threshold value of Value in HSV for finger detection. We apply the labeling process to the binarized image for saving the fingertip area. In the rectangle search, the change of the Hue histogram in the rectangle is used to extract the lunula and the edge of the nail plate. The rectangle moves upward until it finds an edge. The height of the lunula is calculated using the extracted edge. In order to show the effectiveness of the proposed method, we conducted experiments.
(Keyword)
Nail image processing / Lunula of the nail / HSV color / Edge detection using histgram / Image segmentation
22.
Hisaki Omae, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Detect Presence or Absence of Learning Understanding Using Center Cumulative Frequency Comparison Method and Multistage ICA, Proc. of SAMCON 2019, TT9-4-1-TT9-4-4, Chiba, Mar. 2019.
(Summary)
This paper proposes a method to detect the presence and absence of learning understanding using center cumulative frequency comparison (CCFC) method and multistage independent component analysis (ICA). The proposed method consists of four stages: electroencephalogram (EEG) measurement, EEG preprocessing, EEG feature extraction and EEG pattern classification. In the EEG measurement, the EEG signals are measured using a simple electroencephalograph. The EEG preprocessing consists of two phases: eye blink artifacts detection and eye blink artifacts removal. The EEG feature extraction consists of two phases: frequency analysis and band division based on the rhythm of brain activities. In the EEG pattern classification, k-nearest neighbor (k-NN) is used to classify EEG patterns on the basis of band division results. In order to show the effectiveness of the proposed method, this paper conducted three cases. The experimental results suggest that split of the discrimination model using the proposed method is relatively effective when detecting the presence and absence of learning understanding using EEG.
Mana Sasaoka, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Estimate Pressing Positions of Guitar String for Guitar Solo Skill Acquisition, Proc. of SAMCON 2019, Chiba, Mar. 2019.
(Summary)
This paper proposes a method to estimate string pressing positions of the solo part of the lead guitar using image processing technology. The proposed method consists of four phases: image rotation, creation of standard fingerboard image, guitar position detection, and estimation of the string pressing positions. The image rotation method consists of line detection and rotation of the image. The creation of standard fingerboard image consists of detection of frets and intersection detection of frets and strings. In the guitar position detection, the coordinates of the color marker which attach to the bridge and nut are detected. The estimation of the string pressing positions consists of deformation of the standard fingerboard image and detecting finger positions. In order to show the effectiveness of the proposed method, we conduct experiments using videos playing the guitar. The experimental results suggest that the estimation of the string pressing positions in this proposed method is relatively effective.
(Keyword)
image processing / Hough transform / HSV color / Guitar solo skill / Human support
24.
Shun Yamamoto, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Recognition of Aerial Input Numerals by Leap Motion and CNN, Proc. of 2018 SCIS&ISIS, 1189-1192, Toyama, Dec. 2018.
(Summary)
As information technology has advanced in recent years, services which include personal authentication systems such as ATM are increasing. Current main personal authentication systems include IC cards, passwords, and biometrics authentication such as fingerprint authentication.However, there are several problems in these systems.Therefore, better systems are needed.As such systems, we propose a method to write numerals inthe air using the Leap motion and to carry out personal authentication from such aerial handwriting data. We try to identify numerals 0 to 9 as a previous stage. After applying some pre-processing to inputs, learning and identification are carried out using CNN which is a method of machine learning. As a result, average identification accuracy rate was 93.4%. From this result, it is suggested that input numerals in the air can be identified and there is a possibility to construct a new personal authentication system.
Ryohei Shioji, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Personal Authentication and Hand Motion Recognition Based on Wrist EMG Analysis by a Convolutional Neural Network, Proceedings of 2018 Joint 10th International Conference on Soft Computing and Intelligent Systems and 19th International Symposium on Advanced Intelligent Systems, 1184-1188, Toyama, Dec. 2018.
(Summary)
Recent years, biological signals have attractedmuch attention as a tool of human interface. Electromyogram(EMG) has been used in a variety of situations in particular.Generally EMG of muscular volume arms or shoulders has been measured in many cases. In addition, expensive wet typesensors have been often used. However, they are inconvenientand high-cost. On the other hand, in hand motion recognitionand personal authentication using wrist EMG, we haveobtained good results. However, there has been no way toestablish them at the same time. Therefore, in this paper wemeasure EMG by attaching dry type sensors to wrist, andcarry out hand motion recognition and personal authentication.The conventional method on hand motion recognition usedEMG of movement Japanese Janken. The average accuracywas 92.9%. The conventional method on personalauthentication used only "paper" of Japanese Janken. Theaverage accuracy was 96.7%. We used a Convolutional NeuralNetwork (CNN) for learning and identification. In theproposed method, we try to carry out hand motion recognitionand personal authentication at the same time. We use a multiinputand multi-output models of CNN. The average accuracyof hand motion recognition is 94.5%. The average accuracy of personal authentication is 94.57%.
(Keyword)
electromyogram activity / Deep learning / Wrist EMG
Ryohei Shioji, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Personal Authentication and Hand Motion Recognition Based on Wrist EMG Analysis by a wide Residual Network, Proc. of 2018 Annual Conference on Engineering and Applied Science, 38-48, Osaka, Nov. 2018.
(Summary)
Recent years, biological signals have attracted much attention as a tool of human interface. Electromyogram (EMG) has been used in a variety of situations in particular. Generally, EMG in muscular volume of arms or shoulders has been measured in many cases. In addition, expensive wet type sensors have been often used. However, they are inconvenient and high-cost. On the other hand, in hand motion recognition and personal authentication using wrist EMG, we have obtained good results. However, accuracy is poor when hand motion recognition and personal authentication are carried out at the same time. For the above reasons, we carry out hand motion recognition and personal authentication at the same time, and try to obtain higher accuracy than the previous research. The conventional method used EMG of movement Japanese Janken (Fig.1). We use a multi-input and multi-output model of a Convolutional Neural Network (CNN). The average accuracy of hand motion recognition is 94.6%. The average accuracy of personal authentication is 95.0%. In this paper, we use a Wide Residual Network (WRN). The average accuracy of hand motion recognition is 97.8%. The average accuracy of personal authentication is 98.4%. In future work, we aim to improve accuracy by adjusting WRN parameters. In addition, we prepare a class not belonging to any class in multi-class classification.
(Keyword)
electromyogram activity / Deep learning / Wrist EMG
27.
Shion Morikawa, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Personal Authentication by Lips EMG Using Dry Electrode andCNN, Proc. of 2018 International Conference on Internet of Things and Intelligence System (IOTAIS), 180-183, Denpasar, Nov. 2018.
(Summary)
As an alternative to voice, sign language and artificial larynx can be used. However, there are disadvantages where they require a long-term training and are expensive. Therefore, researches on detection of utterance by electromyography (EMG) analysis around the lips have beenconducted. On the one hand, it is necessary to construct a personal authentication system to identify speakers. The electrode used in this paper is 2 electrodes sensor, which is small in size and a dry type. Three sensors are attached in the orbicularis muscle, the zygomatic major muscle, and the depressor angle oris muscle which can acquire myoelectric information necessary for identification in Japanese vowelutterance. EMG signals are measured using P-EMG plus. In order to eliminate noises, signal cutting is carried out before and after the central point of the acquired raw data. Furthermore, EMG data are divided to increase the number of data while overlapping. These are named DATA 1. A Hamming window is then applied for them, and the amplitude values of the power spectra are calculated by fast Fouriertransform. Automatic verification and elimination of noise parts by quartile method were carried out. In order to reconstruct signals after noise elimination, the inverse Fourier transform is carried out and then a inverse Hamming window is applied. These are named DATA 2. Learning identification is carried out using a convolutional neuralnetwork. A large difference was found in accuracy dependingon the data set created separately by measurement date. Therefore, it was found that intra-individual variation by each subject was large. In the future, it is necessary to further improve the data and to reduce individual variation within each subject.
Shun Yamamoto, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Authentication of Aerial Input Numerals by Leap Motion and CNN, Proc. of 2018 International Conference on Internet of Things and Intelligence System (IOTAIS), 189-193, Denpasar, Nov. 2018.
(Summary)
As information technology has advanced in recent years, services which include personal authentication systems such as ATM are increasing. Current main personal authentication systems include IC cards, passwords, and biometrics authentication such as fingerprint authentication. However, there are several problems in these systems.Therefore, better systems are needed. As such systems, we propose a method to write numerals in the air using the Leap motion and to carry out personal authentication from such aerial handwriting data. We try toauthenticate numerals 0 to 9 which are written by three subjects.After applying some pre-processing to inputs, learning and identification are carried out using CNN which is a method of machine learning. As a result, average identification accuracy was 90.3%. From this result, it is suggested that input numerals in the air can be authenticated and there is a possibility to construct a new personal authentication system.
Ryohei Shioji, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Personal Authentication and Hand Motion Recognition Based on Wrist EMG Analysis by a Convolutional Neural Network, Proc. of 2018 International Conference on Internet of Things and Intelligence System (IOTAIS), 184-188, Denpasar, Nov. 2018.
(Summary)
Recent years, EMG has attracted much attention as a tool of human interface. In hand motion recognition and personal authentication using wrist EMG, we have obtained good results. However, there has been no way to establish them at the same time. Therefore, in this paper we measure EMG by attaching dry type sensors to wrist, and carry out hand motionrecognition and personal authentication. The conventional method used EMG of movement Japanese Janken. We use a multi-input and multi-output model of a Convolutional Neural Network (CNN). The average accuracy of hand motion recognition is 94.5%. The average accuracy of personalauthentication is 94.6%. In the conventional method, personalauthentication was classified into two classes. However, we carry out multi-class classification in the proposed method. In feature extraction, we obtain 128×8 input data from the measuring unit. Then, a filter size of the convolution layers is 3×3. CNN does not contain pooling layers in this paper. In the proposed method, the average accuracy of hand motion recognition is 94.6%. The average accuracy of personal authentication is 95.0%.
(Keyword)
electromyogram activity / Deep learning / Wrist EMG
Kazuki Shimamoto, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Extract a Nail Half Moon for Accumulation Stress Evaluation, Proceedings of the 6th IIAE International Conference on Intelligent Systems and Image Processing 2018, 289-292, Matsue, Sep. 2018.
(Summary)
In this paper, we propose a method to extract a nail half moon for accumulation stress evaluation. The proposed method consists of three phases; nail half moon area extraction, HSV color system conversion and nail half moon outline detection. Trimming of the nail half moon area is carried out with fixed parameters because the shooting environment of the nail is fixed. In the HSV color system conversion, the value of hue is used to binarize the nail half moon part and the nail plate part. The outline of the nail half moon is detected using labeling a region that has the maximum value in the nail image. In order to show the effectiveness of the proposed method, we conduct experiments.
(Keyword)
electroencephalogram / nail image procession / accumulation stress / HSV color system
Hisaki Omae, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Check whether Human Understands Contents of Learning Using Electroencephalogram, Proceedings of the 6th IIAE International Conference on Intelligent Systems and Image Processing 2018, 283-288, Matsue, Sep. 2018.
(Summary)
This paper proposes a method to check whether human understands contents of learning using electroencephalogram (EEG). The proposed method consists of three stages; EEG measurement, EEG feature extraction and EEG pattern classification. The EEG feature extraction method (stage) consists of three phases; frequency analysis, band division based on rhythm of brain activities and principal component extraction. The frequency analysis converts an original waveform of EEG into frequency components (bands). In the band division, frequency components were divided into five rhythms ( , Low-, High- , Low- and High- waves). In the principal component extraction, we calculate the first and the second principal components using the principal component analysis (PCA). In the EEG pattern classification, support vector machine (SVM) is used to classify EEG patterns based on PCA results. In order to show the effectiveness of the proposed method, we conduct experiments using real EEG data. The experimental results suggest that the most important sensing position to record EEG is F7 when checking whether he/she understands contents of learning using EEG.
(Keyword)
electroencephalogram / understanding / electroencephalogram / support vector machine
32.
Shin-ichi Ito, Momoyo Ito, Shoichiro Fujisawa and Minoru Fukumi : An Electroencephalogram Analysis Method to Classify Answers of Questions Using Deep Neural Networks, International Conference on Information Technology and Computer Science, 16, Toronto, Aug. 2018.
(Summary)
This paper proposes a method to classify answers of questions during human communications based on an electroencephalogram (EEG) analysis technique. The proposed method consists of three phases; EEG measurement, EEG feature extraction and answers classification. In the EEG measurement, this paper employs the simple electroencephalograph. The EEG signals of frontal lobe are recorded. The EEG features are calculated by normalizing the EEG signals and using deep neural networks. The answers of questions classify by analysing the EEG features using support vector machine. In order to show the effectiveness of the proposed method, we conducted experiments using real EEG data. In the experimental results, we confirm that mean of the recognition accuracy was 93.5%. These results suggest that the answers of questions during human communications can be classified using the EEG signals and their analysis, and that the EEG analysis technique using the deep neural networks and the support vector machine is suitable for extracting the EEG features and classifying the EEG features.
(Keyword)
electroencephalogram / answer of questions / deep neural networks / support vector machine
33.
Ani Firdaus Mohammad, Kamat Rahayu Seri, Minoru Fukumi and Minhat Mohamad : Development of Decision Support System for driving Condition Based on Driving Fatigue Problem, Proc. of Innovative Research and Industrial Dialogue'18, 1-2, Melaka, Jul. 2018.
(Summary)
Driving fatigue has been identified as one of the top contributor factor to the road accidents among the Malaysian road user. The aims of this study was to develop a decision support system for improving the driving fatigue problem. The decision support system provide systematic analysis and solutions to minimize the risk and the number of accidents associated with driving fatigue. Six main components involved as the pillar in the development of decision support system. The decision support system is an essential system to analyze the risk factors that would contribute significantly to driving fatigue associated with driving activity.
(Keyword)
Driving Fatigue / Decision Support System / Road accident / Fuzzy logic
34.
Teruaki Ito, Kamat Rahayu Seri, Minoru Fukumi and Ani Hirdaus Muhammad : Development of working-posture monitoring system for ergonomic manufacturing work environment, Advances in Transdisciplinary Engineering, Vol.7, 1112-1121, Jul. 2018.
(Summary)
Ergonomic consideration on workers in manufacturing work has been attracted attention by industries. Physical safety and mental stability, which would be offered to workers by a well-designed ergonomic work environment, could not only provide job satisfaction to the workers, but also could enhance productivity in manufacturing work. This study develops a prototype monitoring system of ergonomic working posture in a manufacturing work environment. The methodology of this monitoring system is based on the experimental observation using an experimental tool of X-box cameras controlled by XAMPP software to monitor the posture of workers. As for the ergonomic environment measurement, the system installs the five types of instrumental sensors, which include sound, light, temperature, vibration and indoor air quality. Capturing the body posture of subjects, the system measures the frequency of body bending, the angle of body bending and the time period of the same position kept by the body. Potentially useful several parameters are used in the experiments of this study. These parameters include temperature, light, vibration, body posture, indoor air quality and noise. The parameters are captured in analogue signals, which can be converted to digital signals by a signal converter. The analysis of worker posture on RULA (Right Upper Limb Assessment) was also conducted by using several software tools. Reviewing the experimental results using the monitoring system in a manufacturing industry at welding assembly section, this paper shows the feasibility of the proposed system.
Kamat Rahayu Seri, Ani Firdaus Mohamad, Hadi Aisyah Abd Nur, Rayme Syafiqah Nur, Momoyo Ito and Minoru Fukumi : Redesign the Material Handling System by Using Ergonomic Approaches to Reduce Back Pain Risk, International Conference on Kansei Engineering & Emotion Research 2018, 1-11, Kuching, Malaysia, Mar. 2018.
(Summary)
This paper is about the redesign material handling device through the study of psychological and biomechanical factors that related the activities push and pull trolley. The project has been conducted in the automotive industry. The data about the problems encountered through a questionnaire among 14 workers at Receiv-ing Area, since it has the highest number of ergonomic risk factors. The result shows the main problem is the existing trolley design seems not suitable and giv-ing musculoskeletal disorder(MSD) of the workers especially back pain. There-fore, the aim of this paper is redesigns the material handling system by using er-gonomic approaches to reduce back pain. The Tekscan software will be used for evaluating hand pressure distribution force on the existing and alternatives trolley. The result obtained is used as a benchmark for the concept of trolley to be rede-signed. The Rapid Upper Limb Assessment (RULA) was applied to analyses the posture of workers on the existing and alternative trolley and the redesign trolley to be proposed by ABC company. Based on the RULA analysis, the redesigned trolley has improved the posture of the workers. Hence, this study concludes that considering ergonomic features for the redesigned trolley contributed to safe body posture.
Ani Firdaus Mohamad, Minoru Fukumi and Kamat Rahayu Deri : Development of Decision Support System for Improving Driving Fatigue Problem among Road Users, 4th International Forum on Advanced Technologies (IFAT2018), 1-3, Tokushima, Mar. 2018.
(Summary)
The aim of this study is to develop an ergonomic vehicle model (EVM) and decision support system (DSSfDF) model for improving the driving fatigue problem among the road users. The ergonomic vehicle model use to capture the user information data, and acts as the database storage to store all the input data and information. While the decision support system provide a systematic analysis and solution to minimize the risk and the number of accidents associated with driving fatigue. There are 6 main components as the pillars for the development of EVM and DSSfDF model; ergonomics evaluation tools, graphical user interface (GUI), ergonomics database, working memory, inference engine, and knowledge base. Both models are essential system and reliable advisory tool for providing analy-sis on risk factors that contribute significantly to driving fatigue, and providing solutions and recommendation to the problem related to driving fatigue. Further analysis and validation is required in future to get the reliable system before being commercialize.
37.
Kamat Rahayu Seri, Ani Firdaus Mohamad, Ghazali Athira, Shamsudin Syamami, Momoyo Ito and Minoru Fukumi : Mathematical Modelling of Biomechanics Factors for Push Activities in Manufacturing Industry, Symposium on Intelligent Manufacturing and Mechatornics 2018, 3-14, Pekan, Malaysia, Jan. 2018.
(Summary)
In manufacturing industries, many working tasks require their workers to perform the works in push-pull activities. The workers need to push or pull the tool or material handling in a long distance in to a workplace and performing these activities continuously throughout the working hours, may lead to an early initiation of musculoskeletal disorders (MSDs) symptoms as workers developed muscle fatigue particularly concerning the hand muscles. Grip strength is the force applied by the hand to pull objects and is a part of hand strength. This paper is about the mathematical model of biomechanical factors that contributes to fa-tigue while worker involved on the push activities in manufacturing industry. The experimental was conducted by using Tekscan system to evaluate the muscle fa-tigue and hand grip pressure force while workers performing pushing excessive loads. The input parameters were time exposure, hand side and body mass index (BMI); while the output responses are muscle fatigue (voltage), hand grip pres-sure force (left hand), and hand grip pressure force (right hand). An important parameter that affects the output response is also identified. The finding result from mathematical model for both factors, show that the muscle fatigue was in-fluenced by time exposure, hand side, BMI, and interaction between hand side and BMI; while hand grip pressure force was influenced by time exposure, hand side, BMI, interaction between time exposure and hand side, interaction between time exposure and BMI, and interaction between hand side and BMI.
Ani Firdaus Mohamad, Kamat Rahayu Seri, Minoru Fukumi, Momoyo Ito, Minhat Mohamad and Rayme Syafiqah Nur : Development of Ergonomic Vehicle Model and Decision Support for Driving Fatigue, Symposium on Intelligent Manufacturing and Mechatornics 2018, 355-369, Pekan, Malaysia, Jan. 2018.
(Summary)
Driving fatigue has been recognized as one of the significant contrib-utor factor to the road accidents and fatalities in Malaysia. The aim of this study was to develop an ergonomic vehicle model (EVM) and decision support sys-tem (DSSfDF) model for improving the driving fatigue problem among the road users. The ergonomic vehicle model use to capture the user information data, and acts as the database storage to store all the input data and information. While the decision support system provide a systematic analysis and solution to minimize the risk and the number of accidents associated with driving fatigue. There are 6 main components as the pillars for the development of EVM and DSSfDF model; ergonomics evaluation tools, graphical user interface (GUI), ergonomics database, working memory, inference engine, and knowledge base. Both models are essential system and reliable advisory tool for providing analy-sis on risk factors that contribute significantly to driving fatigue, and providing solutions and recommendation to the problem related to driving fatigue. Further analysis and validation is required in future to get the reliable system before be-ing commercialize.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : An Electroencephalogram Analysis Method to Detect Preference Using Gray Association Degree, The proceedings of International Conference on Electronics, Information, and Communication 2018, 304-305, Honolulu, Jan. 2018.
(Summary)
This paper introduces an electroencephalogram (EEG) analysis method to detect human preference. The proposed method consists of three phases; EEG recording, EEG feature extraction and preference detection. In EEG recording, we employ the simple electroencephalograph. The measurement position to record the EEG is left frontal lobe (FP1). The gray association degree is used to extract the EEG feature. The support vector machine is used to detect human preference on sounds listened to. In order to show the effectiveness of the proposed method, we conduct the experiments. In the experimental results, the mean of the accuracy rate of the favorite sound detection was higher than 88%.
(Keyword)
electroencephalogram / preference / gray association degree / support vector machine
Ani Firdaus Mohammad, Mahmood Hasrulnizam Wan Wan, Minhat Mohamad, Kamat Rahayu Seri and Minoru Fukumi : Development of Driving Fatigue Strain Index to Analyze Risk Levels of Driving Activity, Proceedings of 2017 International Conference on Electrical, Electronics and System Engineering (ICEESE 2017), 95-99, Kanazawa, Nov. 2017.
(Summary)
Driving activity has become more important as this medium being practically, faster and cheaper in connecting human from one to another place. However, driving activity can cause disaster or death to human in daily life as they get fatigued while driving. Driver fatigue is a top contributor to the road crashes. The primary objective of this study was to develop a driving fatigue strain index (DFSI), to quantify the risk levels caused by driving activity, and proposed solution in minimizing the number of road accidents caused by driving fatigue. The development of DFSI is based on risk factors associated with driving activity such as muscle activity, heart rate, hand grip force, seat pressure distribution, whole-body vibration, and driving duration. All risk factors were assigned with multipliers, and the DFSI was the output or result of those multipliers. The development of DFSI is essential to analyze the risk factors that would contribute significantly to discomfort and fatigue associated with driving. Besides, in the future this index has a capability to recommend alternative solutions to minimize fatigue while driving.
Takashi Higasa, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Specify a Region of Character String in Augmented Reality, Proc. of the 5th IIAE International Conference on Intelligent Systems and Image Processing 2017, 25-30, Honolulu, Sep. 2017.
(Summary)
This paper proposes a method to input characters and/or character string in an augmented reality using gesture motion. The proposed method detects the region of character string using gesture motion. It consists of five phases; template generation, skin detection, hand region detection, gesture motion extraction and designation of character string region. The template image consists of two fingers because a gesture is to take hold the tips of the first and second fingers. In the skin detection, we extract the skin color on the basis of values in saturation by using threshold processing. The hand region is detected by calculating areas and detecting the area with the maximum value as a hand. The gesture motion is extracted using template matching. In order to show the effectiveness of the proposed method, we conduct experiments.
(Keyword)
Augmented Reality
42.
Kohei Nakanishi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Biometrics Authentication of Aerial Handwritten Signature Using a Convolutional Neural Network, Proc. of the 5th IIAE International Conference on Intelligent Systems and Image Processing 2017, 19-24, Honolulu, Sep. 2017.
(Summary)
Recent years, biometrics authentication is receivingattention by development of information society. In thispaper, we propose a personal authentication system, whichuses behavior characteristics among biometrics. We focuson aerial handwritten signature, because it is difficult toforge it, and there is no risk of loss. In this paper, wemeasure signatures using Leap Motion Controller. It canmeasure three dimensional space coordinates with highaccuracy. We divide signature data into three axialdirections of coordinates XYZ in order to use them asone-dimensional data. We carry out preprocessing tosignature data and normalize them. Next, we use deeplearning based on a convolutional neural network forfeature extraction and identification. In this experiment, weprepare data obtained from six subjects. We obtain genuinedata of one subject. The remaining five subjects are used tocreate forgery data. We classify signature data into twoclasses. We conduct deep learning in which convolutionalneural network carries out 10,000 cycles learning in onetrial. We carry out this trial 5 times and evaluate meanaccuracy by cross validation for two types of genuine data.The average discrimination accuracy of this experiment are97.0 % and 95.9%. In addition, the false rejection rates are9.6% and 19.2%. The false acceptance rate are 0.8% and0.1%.
(Keyword)
neural network / Deep learning
43.
Ryohei Shioji, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Personal Authentication Based on Wrist EMG Analysis by a Convolutional Neural Network, Proc. of the 5th IIAE International Conference on Intelligent Systems and Image Processing 2017, 12-18, Honolulu, Sep. 2017.
(Summary)
Recent years, biological signals have attracted muchattention as a tool of human interface. Electromyogram(EMG) has been used in a variety of situations in particular.We measure EMG of arms or shoulders in many cases. Inaddition, we often use expensive wet type sensors. However,they are inconvenient and high-cost. On the one hand, therehave been few works of personal authentication using EMG.Therefore, in this paper we measure EMG by attaching drytype sensors to wrist, and carry out personal authentication.The conventional method in this paper is divided into threeunits such as a measuring, a feature extraction, and adiscrimination units. We measure EMG signals with eightdry type sensors on the wrist. After that, we identify amotion opening our hands. We use a convolutional neuralnetwork (CNN) to learning and authentication. Wecollected 40 data for each subject. The average accuracy oftwo-class separation was 94.9 % by CNN. In addition to theconventional method, the proposed method in this paperpreprocesses the data. Large noise was removed using ahigh path filter. By this preprocessing, identificationaccuracy (Two-class classification using CNN) improvedby 1.5%. The true acceptance rate improved by 7.2%, andthe false acceptance rate improved by 0.0067%.
(Keyword)
electromyogram activity / Deep learning
44.
Ryousuke Takabatake, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Japanese Vowels Recognition Using Linear Discriminant Analysis and Surface Electromyogram Measured with Bipolar Dry Type Sensors, Proc. of the 5th IIAE International Conference on Intelligent Systems and Image Processing 2017, 5-11, Honolulu, Sep. 2017.
(Summary)
This paper proposes a Japanese vowels recognition method using surface electromyogram (EMG). First, 3 sensors are used to measure surface EMG data at orbicularis oris muscle, zygomatic muscle and depressor angle oris muscle. Next, Fast Fourier Transform (FFT) is applied to all measurement data to calculate power spectra. Linear Discriminant Analysis (LDA) is then used for power spectra of 3 channels and reduce their dimension to 4. Finally, theresult of LDA is recognized by Support Vector Machine (SVM). In experiments, it is assumed that mounting sensors to face, measuring EMG, and demounting them are 1 trial. A subject utters 5 Japanese vowels 3 times. Among 3 trials data, 2 trials data are used to make templates and the remaining are used for test. The subject is a man in twenties. As a result, we obtained 62.3% average recognition accuracy. This result shows the proposed method is better about 2 times than the previous method.
(Keyword)
electromyogram activity / SVM
45.
Ani Firdaus Mohammad, Hamid Hafiz A Mohd, Shamsuddin Syamimi, Minoru Fukumi and Teruaki Ito : A Study of Biomechanical Factor for Driver Fatigue using Regression Model, Proceedings of International Conference on Design and Concurrent Engineering Conference 2017 & Manufacturing Systems Conference 2017, Vol.17, No.205, 9-1-9-10, Osaka, Sep. 2017.
46.
Kamat Rahayu Seri, Hamid Hafiz A Mohd, Shamsuddin Syamimi, Minoru Fukumi, Teruaki Ito and Husain Kalthom : Ergonomics Study Of Working Postures In Demould Process At Aerospace Manufacturing, Proceedings of International Conference on Design and Concurrent Engineering Conference 2017 & Manufacturing Systems Conference 2017, Vol.17, No.205, 8-1-8-10, Osaka, Sep. 2017.
47.
Ryohei Shioji, Daiki Hiraoka, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Personal Authentication Based on Wrist EMG Analysis by a Convolutional Neural Network, Proceeding of International Conference on Advanced Technology & Sciences (ICAT'Rome), 335-340, Rome, Nov. 2016.
(Summary)
In this paper we measure EMG by attaching dry type sensors to wrist, and carry out personal authentication. The proposed method in this paper is divided into three units such as a measuring, a feature extraction, a discrimination units. We measure EMG signals with eight dry sensors on the wrist. After that, we identify a motion opening our hands. We use a convolutional neural network (CNN) to learning and authentication. In addition, we try to use a multilayer perceptron for comparison. Experiments are conducted in two patterns. At first, we carry out two-class classification (the subject and the others). The second is multi-class classification in which the number of subjects is 8 people. We collected 40 data for each subject. The average accuracy of two-class classification was 89.4 % by the multilayer perceptron. That was 94.9 % by CNN. On the other hand, the average accuracy of multi-class classification was 41.2 % by the multilayer perceptron. That was 70.3 % by CNN. In future work, we will improve classification accuracy for two-class and multi-class classification. Futhermore, we try to identify multiple motions.
48.
Kohei Nakanishi, Daiki HIraoka, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Biometric Authentication of Aerial Handwritten Signature Using a Convolutional Neural Network, Proceeding of International Conference on Advanced Technology & Sciences (ICAT'Rome), 329-334, Rome, Nov. 2016.
(Summary)
In this paper, we propose a personal authentication system which uses behavioral characteristics among biometrics. We focus on aerial handwritten signature, because it is difficult to forge it, and there is no risk of loss. In this paper, we measured signatures using Leap Motion Controller. It can measure three dimensional space coordinates with high accuracy. We divide signature data into three axial directions of coordinates XYZ in order to use them as one dimensional data. We carry out preprocessing to signature data and normalize them. Next, we use deep learning based on a convolutional neural network for feature extraction and identification. Generally, it is necessary to learn a large number of data for deep learning. However, it is difficult to gather many learning data of forgery. Therefore, we transform learning data of forgery at the preprocessing, which generates many pseudo learning data of forgery. In this experiment, we prepare genuine data of nine persons and forgery data of each person. We classify signature data into ten classes. We conduct deep learning in which CNN carries out 10,000 cycles learning in one trial. We conduct this trial 1,000 times. The average discrimination accuracy of this experiment is 98.1 %. This result is better in accuracy than a related research. From the above, we think that the proposed method is useful for the aerial handwritten signature.
49.
Daiki Hiraoka, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Japanese Janken Recognition Based on Wrist EMG Analysis by CNN and SVM, Proceeding of International Conference on Advanced Technology & Sciences (ICAT'Rome), 323-328, Rome, Nov. 2016.
(Summary)
In this paper, we propose a method which can discriminate hand motions. We measure an electromyogram of wrist by using 8 dry type sensors. We focus on four motions, such as ``Rock-Scissors-Paper'' and ``Neutral''. ``Neutral'' is a state that does not do anything. The proposed method extracts features of EMG by a convolutional neural network (CNN) and discriminate the motions by a support vector machine (SVM). In the CNN, we reduced the full connection layer by adding a convolution layer which has the same size filter of a feature map. CNN has an input layer, 6 convolutional layers, a pooling layer and a full connection layer. We conducted experiments with seven subjects. An average discrimination accuracy of the proposed method was 92.2 %. In the previous method, the discrimination accuracy rate was 76.9%. Therefore, the proposed method is better in accuracy than the previous method. In future work, we will discriminate more detailed hand motions.
50.
Taiki Nonoguchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Detect and Track Mosquitoes Using Orientation Code Matching and Particle Filter, Proceeding of the 2016 International Conference on Electrical Engineering, ID90204-1-6, Naha, Jul. 2016.
(Summary)
In this paper, we propose a method to detect and track mosquito by using orientation code, labeling, multi-templates matching and particle filter. The mosquito detection method employs an edge detection technique based on orientation code, a labeling technique to detect candidate areas of mosquitoes and multi-templates matching to detect mosquitoes. The particle filter is used to track the mosquitoes. The likelihood in the particle filter is calculated on the basis of the results of the multi-templates matching. In order to show the effectiveness of the proposed method, we conduct experiments using real image data. From experimental results, detection and tracking accuracies were 83.9% and 70.6%, respectively.
51.
Shunsuke Takata, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : A Basic Study for Driver State Estimation Based on Time Series Data Analysis, Proceeding of the 2016 International Conference on Electrical Engineering, ID90096-1-6, Naha, Jul. 2016.
52.
Ryosuke Takabatake, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Vowel sound recognition using electromyogram with dry sensors, Proceeding of the 2016 International Conference on Electrical Engineering, ID90049-1-5, Naha, Jul. 2016.
(Summary)
This paper proposes a vowels recognition method using electromyogram (EMG). 3 sensors are used to measure EMG data at orbicularis oris muscle, zygomatic muscle and depressor angle oris muscle. Fast Fourier Transform (FFT) is applied to all measurement data and then principal component analysis (PCA) is used for phase spectra and power spectra of 3 channels.Finally, the result of PCA is recognized by k-nearest neighbor. In experiments, it is assumed that mounting sensors to face,measuring EMG, and demounting them are 1 trial. A subject utters 5 Japanese vowels 3 times. Among 3 trials data, 2 trialsdata are used to make templates and the remaining are used for test. The subject is a man in twenties. As a result, we obtained33% average recognition accuracy. As future tasks, the authors think it is necessary to change methods of decision of datarange, feature extraction, recognition and so on.
53.
Takuma Ogawa, Daiki Hiraoka, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Improvement in Detection of Abandoned Object by Pan-tilt Camera, Proceeding of the 2016 8th International COnference on Knowlodge and Smart Technology, 152-157, Chiang Mai, Feb. 2016.
(Summary)
We propose a method which detects abandoned objects on online by using pan-tilt camera. Above all, we improve problems of the previous method which is based on ST-Patch features and human detection. We make extended ST-Patch features for solving the problem of ST-Patch features. We improve human detection by using deep learning which is based on a convolutional neural network. We conducted preliminary experiments to verify a method of pooling, and then we decided to use Max pooling because its detection accuracy is better than that of Ave booling. We conducted experiments in five situations to verify usefulness of the proposed method. If the proposed method finds an abandoned object, it saves the object image. We define the abandoned object as an object which human does not subsist near. We could detect the abandoned object in each situation. However, we conducted experiments of the proposed method only in a room. We need to conduct experiments in a wide area to find new problem.
Daiki Hiraoka, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Japanese Janken Recognition by Support Vector Machine Based on Electromyogram of Wrist, Proceeding of the 2016 8th International COnference on Knowlodge and Smart Technology, 114-119, Chiang Mai, Feb. 2016.
(Summary)
In this study, we propose a method which can discriminate hand motions. We measured an electromyogram of wrist by using 8 dry type sensors. We focused on four motions, such as ``Rock-Scissors-Paper'' and ``Neutral''. ``Neutral'' is a state that does not do anything. In the proposed method, we apply fast Fourier transformation (FFT) to measured EMG data, and then remove hum noise. Next, we combine values of sensors based on a gaussian function. In this gaussian function, variance and mean are 0.2 and 0, respectively. After that, we apply normalization by linear transformation to the values. Subsequently, we resize the values into range from -1 to 1. Finally, support vector machine (SVM) conducts learning and discrimination. We conducted experiments in three subjects. Discrimination accuracy of the proposed method for three subjects was 96.9%, 95.3%, 92.2%, respectively. Therefore, we think that the gaussian function is robust to difference of sensor position because this function combines both adjacent channels.
Shu Tamura, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Method to Evaluate Similarity of Music by Music Features, 42st Annual Conference of the IEEE Industrial Electronics Society (IECON 2015), 2574-2577, Yokohama, Nov. 2015.
(Summary)
This paper proposes a method to evaluate similarity of music by music features. A music feature extraction method consists of three phases; chord progression pattern detection, rhythm pattern detection and musical instrument information extraction. The music feature extraction is carried out by using frequency analysis. In the chord progression pattern, we employ three evaluation criteria. In the rhythm pattern detection, we evaluate beat per minutes(BPM) values. In the musical instrument information, we confirm results of musical instrument informationextraction visually. In order to show the effectiveness of the proposed method, we conduct computer simulations of music features extraction.
Daiki Hiraoka, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Hand Motion Discrimination by Support Vector Machine Based on Electromyography of Wrist, Proceeding of the 2015 International Conference on Engineering and Applied Science, 358-366, Sapporo, Jul. 2015.
(Summary)
In this study, we propose a method which can discriminate hand motions. We measured an electromyography of wrist by using 8 dry type sensord. We focus on four motions, such as "Rock-Scissors-paper" and "neutral". "neutral" is a state that does not do anything. In the proposed method, we apply fast Fourier transformation (FFT) to the measured EMG data, and then remove hum noise which are caused by AC power source. After that, we apply normalization by linear transformation to FFT spectra.. Finally, SVM learns 4096 data which are all data of 8 channels. an experiment of this study has a specific flow. Star of the experiment is to attach sensors, and the end of experiment is to detach the sensors. We regard this flow as 1 trial. In each trial, a subject conducted each motion 10 times. After the experiment, data of each motion are picked out from measured data, and then these data were used for learning and discrimination. Discrimination accuracy was 98.4%. However the subject of experiment was 1 person. Therefore, in future work we need to increase the number of subjects to validate versatility proposed method. In addition, our proposed method is offline now. Therefore, we will implement this method on online.
57.
Takahide Funabashi, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : On-line Recognition of Finger Motions Using Wrist EMG and Simple-PCA, Proc. of Asian Control Conference 2015 (ASCC'2015),, 2182-2186, Kota Kinabalu, Jun. 2015.
(Summary)
This paper presents an on-line recognition system of finger motions using wrist EMG (electromyogram) measured by dry-type electrodes attached to wrist and Simple-PCA. The Simple-PCA is an approximated version of principal component analysis and is very fast for eigenvector learning. Target behaviors to be recognized in this paper are four finger motions, which are the Janken (Rock-paper-scissors game) ``rock'', ``scissors'', ``paper'' and ``neutral (non-action)''. We tried to reduce an execution time by using the Simple-PCA in training and recognition, in the viewpoint of implementation of interface which can be utilized in daily life. The computational results show that the present on-line system can achieve high recognition accuracy similar to the conventional system using a neural network classifier.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Method to Detect Impression Evaluation Patterns on Music Listened to Using EEG Analysis Technique, The 10th Asian Control Conference (ASCC 2015), 1848-1853, Kota Kinabalu, May 2015.
(Summary)
In this paper, we propose a method to detect impression evaluation patterns on music listened to using electroencephalogram (EEG) analysis method considering human personality. The proposed method consists of four phases; EEG recordings and EEG feature extraction, personality quantification, feature vector creation to detect the impression evaluation patterns, and impression evaluation patterns detection. The EEG feature is extracted by calculating the time average of the power spectrum of each frequency band at 1 Hz intervals of the EEG. Egogram, Yatabe-Guilford personality inventory and Kretschmer type personality inventory are using for quantifying his/her character. The feature vector to detect the impression evaluation patterns is created by the EEG feature and the results of his/her character quantification. We regard the matching patterns between music and his/her mood as the impression patterns on music listened to. In order to show the effectiveness of the proposed method, we conduct experiments using real EEG data.
Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Feeling Evaluation Detection for Auto-skip Music using EEG Analysis Technique, International workshop on Sensing, Actuation, Motion Control, and Optimization (SAMCON2015), Vol.IS3-2-3, 1-4, Nagoya, Mar. 2015.
(Summary)
This paper proposes a method to detect feeling evaluation patterns on music listened to using electroencephalogram (EEG) analysis method considering human personality. The proposed method consists of four phases; EEG recordings and EEG feature extraction, personality quantification, feature vector creation to detect the feeling evaluation patterns, and feeling evaluation patterns detection. The EEG feature is extracted by calculating the time average of the power spectrum of each frequency band at 1 Hz intervals of the EEG. Egogram, Yatabe-Guilford personality inventory and Kretschmer type personality inventory are using for quantifying his/her personality. The feature vector to detect the feeling evaluation patterns is created by the EEG feature and the results of his/her personality quantification. We regard inclination patterns, which are listening, not listening to the music and feeling borderline case, respectively, as the feeling patterns. In order to show the effectiveness of the proposed method, we conduct experiments using real EEG data.
(Keyword)
decision making / pattern recognition / music / electroencephalogram / individual difference / personality / egogram / YG / KT
60.
Daiki Hiraoka, Minoru Fukumi and Koji Kashihara : Estimation of physical burden in daily living activity by wearable sensors, Proceeding of the 2015 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP), 473-476, Kuala Lumpur, Mar. 2015.
(Summary)
In this paper, we detect daily living activity by using accelerometer, and analyze the relation between the activity and heart rate. The nal purpose is to build a cloud computing system for alerting the physical burden by daily living activity. First, we measure acceleration of daily living activity by using accelerometer and gravity sensor mounted on Android. Next,we apply many lterstovalues measured by using such devices. After that, we detect the activity by using template matching and threshold process. Finally, we calculate a heart rate difference between resting state and moving state, and alert to physical burden.
61.
Nao Tsuzuki, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Detect Uncomfortable Feeling of Listeners by Biological Information, Proceeding of the 2015 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP), 481-484, Kuala Lumpur, Mar. 2015.
(Summary)
In this paper, we propose a method to detect uncomfortable feeling of listeners using biological information analysis techniques. The proposed method supports to hold a conversation for smooth communication. We employ the electroencephalogram (EEG) analysis as the biological information analysis. Independent component analysis (ICA) and fast Fourier transform (FFT) are used to detect specific signals of EEG related to uncomfortable feeling and reduce a noise. In order to show the effectiveness of the proposed method, we conduct experiments using real EEG.
62.
Takuma Ogawa, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Detection of Abandoned Object by Pan-Tilt Camera, Proceeding of the 2015 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP), 278-281, Kuala Lumpur, Mar. 2015.
(Summary)
In this paper, we propose a method of detection for aban- doned objects by a pan-tilt camera. The final purpose is to do online detection of abandoned objects by using the pan-tilt camera. First, we detects object domains by using ST-Patch features from an obtained moving image. We use these fea- tures to efficiently separate moving objects and background. Next, we focus on the object domains. We obtain a picture that was optically expanded by using zoom function. As a re- sult, we can take a clearer picture. After that, we detect these object domains as human or nonhuman by using HOG fea- tures and Real AdaBoost. The HOG features are converted into one-dimensional histogram corresponding to feature val- ues. Detection of human is carried out by Real AdaBoost using the histogram values. The other objects besides human are regarded as abandoned objects in this paper.
63.
Koji Miyai, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Fundamental study for driving scene classification using Bag of Keypoints, Proceeding of the 2015 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP), 342-345, Kuala Lumpur, Mar. 2015.
(Summary)
In this study, we define the scene in front of the vehicle as driving scene and aim at the classification of the driving scenes. Bag of Keypoints (BoK) is a technique often used in image classification. BoK's effectiveness has been shown in the field of object recognition. Then, we have performed classification experiments for driving simulator images by the BoK. We examine the significance of applying the BoK for driving scene classification.
64.
Taito Mori, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Analysis of Driver's Eye-gaze Movements at Near-miss Events, Proceeding of the 2015 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing (NCSP), 330-333, Kuala Lumpur, Mar. 2015.
(Summary)
In this paper, we analyze the relationship between near-miss events and saliency map. To analyze eye-gaze points of the driver for near-miss events which occur in the intersections, we use saliency map which models the human attention mechanism. We use an eye tracking system (faceLAB) to obtain eye-gaze data. We make target intersection time-series driving scene and made saliency map of the scene. Moreover, the saliency map is divided into three levels and we investigate how bicycle and background have saliency in the driving scene. Experimental result shows that the bicycle's saliency is expressed in the middle and high level maps. Next, we carry out experiments using faceLAB data in order to investigate driver's eye-gaze. From the experimental results, subject gaze to similar area when near-miss event did not exist. However, subjects tracked the bicycle when near-miss event occurred, and gazed carefully to the left or right in the intersection after near-miss event.
65.
Takako Ikuno, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Object Search by Template Matching using Genetic Algorithm and Random Search, Proceeding of the 2015 Joint Conference of the International Workshop on Advanced Image Technology (IWAIT) and the International Forum on Medical Imaging in Asia (IFMIA), No.OS.25, 1-4, Tainan, Jan. 2015.
(Summary)
In this paper, we propose a method in which pictures of security cameras are administered automatically. The administered target is lost property. In case of searching objects with security camera, there are infinitely various sizes and orientations of the object to be searched. Therefore we propose an object search method which is adapted to transformation of the object. We use a template matching using Genetic Algorithm (GA) for detection of lost property. Moreover GA is not necessarily suitable for local search problems. Therefore the local search technique using random search is included to improve GA property. Object search in our proposed method is divided into two parts, global and local searches. According to experimental results, in the global search, search accuracy is relatively good at experiments other than the specific images. However the local search fell short of our expectations. In the future, we need to improve the local search.
66.
Stephen Githinji Karungaru, Minoru Fukumi and Kenji Terada : Human Action Recognition using Normalized Cone Histogram Features, Proceedings of the IEEE Symposium Series on Computational Intelligence (CIMSIVP 2014), 12-16, Florida, Dec. 2014.
(Summary)
In this paper, we propose a normalized cone histogram features method to recognize human actions in video clips. The cone features are extracted based not on the center of gravity as is common, but on the head position of the extracted human region. Initially, the head, hands and legs positions are determined. Thereafter, the distances and orientations between the head and the hands and legs are the extracted and employed as the features. The histogram's x-axis represents the orientations and the y-axis the distances. To make the method invariant to human region sizes, the features are normalized using the L2 normalization technique. The classification method used was the perceptron neural network. We conducted experiments using the ucf-sports-actions database to verify the effective ness of our approach. We achieved an accuracy of about 75% on a selected test set.
Yutaka Kameda, Minoru Fukumi and Koji Kashihara : Development of a Healthcare Monitoring System Based on Pulse Wave Analysis, Proceedings of 2014 IEEE 3rd Global Conference on Consumer Electronics (GCCE 2014), 44-46, Tokyo, Oct. 2014.
Takako Ikuno, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Lost Property Detection by Genetic Algorithm with Local Search, Proceeding of The 2nd International Conference on Intelligent Systems and Image Processing 2014, 245-249, Kitakyushu, Sep. 2014.
(Summary)
In this paper, we propose a method in which pictures of security cameras are administered automatically. The administered target is lost property. In case of searching objects with security camera, there are infinitely various sizes and orientations of the object to be searched. Therefore, we propose an object search method which is adapted to transformation of the object. We use a template matching using Genetic Algorithm (GA) for detection of lost property. Moreover, GA is suitable for global search problems, but it is not necessarily suitable for local search problems. Therefore the local search technique is included to improve GA property. Object search in our proposed method is divided into two parts, global search and local search. In the local search, we use a simple random search. According to experimental results, in the global search, search accuracy is relatively good in the almost experiments, but the local search is not so effective in almost experiments. In the future, we need to improve fitness function in the global and local search.
69.
Zhang Peng, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Development of EOG Mouse Using Learning Vector Quantization, Proceeding of the 2nd International Conference on Intelligent Systems and Image Processing 2014, 38-43, Kitakyushu, Sep. 2014.
(Summary)
Recognition of eye motions has attracted more and more attention of researchers all over the world in recent years. In particular studies to make lives more convenient for patients as ALS who cannot move even their muscles expect eye have been actively done. Many kinds of eye motion recognition methods have been proposed, for example, using infrared to track pupil position, or using image processing technologies to find the pupil. However as they use either infrared or cameras, they may have some effect on the eye. Therefore, in our study, we use an EOG method to recognize eye motions: attaching wet disposable electrodes on the patients' face to obtain eye movement signals. Then Learning Vector Quantization algorithm is used to recognize each eye motion. Finally, corresponding recognition results to various mouse operations can be obtained. In our study, we recognized eye motions of rolling eyes upward, downward, rolling left, rolling right, diagonal eye motions which contain rolling up-left, rolling up-right, rolling down-left, rolling down-right ( the angle of the diagonal motion is 45°), blink, blink string of three times motion, left wink and right wink, in total 12 kinds of eye motions. The average recognition accuracy was over 98%. Using this recognition system, we achieved 8 direction cursor movements and double click action, scroll page upward and downward. This study would be used as a means of communication to help those patients as ALS. This system is implemented on a PC as the on-line system.
70.
Momoyo Ito, Kazuhito Sato and Minoru Fukumi : Driver Body Information Analysis with Near-miss Events, Proceedings of AMBIENT 2014: The Fourth International Conference on Ambient Computing, Applications, Services and Technologies, 43-46, Rome, Aug. 2014.
(Summary)
This study specifically examines safety verification behaviors and near-miss events at non-regulated intersection with poor visibility. Assessing the drivers physical information (i.e., eye-gaze movements and face orientation) and the sudden appearance of bicycles that the driver would encounter while approaching a non-regulated intersection, we attempt to analyze the causal relation of workload sensitivity and driving style and the distinctive motion of the safety verification behaviors before and after near-miss events.
71.
Stephen Githinji Karungaru, Kenji Terada and Minoru Fukumi : Vehicle Extraction from Aerial Images Captured using an UAV, Proceedings of the 2014 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing(NCSP'14), 313-316, Honolulu, Hawaii, USA, Mar. 2014.
72.
Akiko Sugiyama, Momoyo Ito, Kazuhito Sato, Shin-ichi Ito and Minoru Fukumi : Analysis of Driving Behavior Caused by Hiyari-Hatto Event Focusing on Head Motion, Proceeding of 2014 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing, 705-708, Honolulu, Mar. 2014.
(Summary)
In this paper, we analyze projected head positions of safety verification on the two-dimensional plane to clarify what head posture changes behavior are contained in the data. In this experiment, we consider the projected coordinate changes due to the presence or absence of hiyari-hatto events, and discuss the relation between the head position changes and the influence of hiyari-hatto event. From the experimental results, the driver becomes cautious after the event, and a change occurs in the safety verification behavior by becoming cautious.
73.
Tomoaki Chika, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Personal authentication system by using Kinect, Proceeding of 2014 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing, 289-292, Honolulu, Mar. 2014.
(Summary)
In this paper, the objective is to propose a method to perform personal authentication from motion information signed in space obtained using the Kinect sensor. Biometrics authentication is one of the hottest authentication systems all over the world. It is high secure compared with other authentication systems. However, it has an issue that is generally expensive. Therefore, we propose an inexpensive biometrics authentication system by using the Kinect sensor. We can obtain hand positions easily because the Kinect sensor can obtain skeleton data quickly. Therefore, it is thought that we can develop a system to track a fingertip.
74.
Daiki Konishi, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Comparison of Poolong Methods in a Deep Neural Network, Proceeding of 2014 RISP International Workshop on Nonlinear Circuits, Communications and Signal Processing, 285-288, Honolulu, Mar. 2014.
(Summary)
The objective of this study is to construct a personal identification system using face with high recognition performance. We define the high recognition performance as being robust to change of surroundings (light, angle, size, and so on). However it is difficult to achieve that with ordinary image processing methods. Therefore we use the Deep Learning method based on Convolutional Neural Network to achieve our objective. In this paper, we compare the pooling methods (max-pooling and average-pooling) as preliminary step of this study. The pooling is one of elements of Convolutional Neural Network. To simplify a problem, we carry out face recognition using face and object images.
75.
Takahiro Horiuchi, Yohei Takeuchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Creation of a Panoramic Image by Genetic Algorithm, Proceeding of 2013 International Conference on System, Process, and Control, 113-116, Kuala Lumpur, Dec. 2013.
(Summary)
The purpose in this paper is to generate a panoramic image without using special equipments such as a panoramic camera. Our proposed method performs image processing with a computer. This paper therefore presents a new panoramic image generation method that solves drawbacks of conventional generating methods. This paper intends to perform high-speed and highly precise generation of a panoramic image. A template matching method using a genetic algorithm (GA) to make a panoramic image is presented. The effectiveness of the present method is demonstrated by measn of computer simulations.
Zhang Peng, Yohei Takeuchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Implementation of EOG Mouse Using Learning Vector Quantization and EOG-feature Based Methods, Proceeding of 2013 International Conference on System, Process, and Control, 98-102, Kuala Lumpur, Dec. 2013.
(Summary)
In this paper we proposed a system that uses EOG-feature based methods and Learning Vector Quantization algorithm to recognize eye motions. According to the recognition results, we use API to control cursor movements. We recognized 12 eye motions. 8 directions motions correspond to 8 directions cursor movement in this system. Using this system we have obtained a high recognition accuracy of eye motions (The average correct detection rate on each subject was 97.8%, 97.6% and 92.7%). This EOG Mouse interface would be used as a means of communication to help those patients as ALS.
Takako Ikuno, Yohei Takeuchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Abandoned Object Detection by Genetic Algorithm with Local Search, Proceeding of 2013 International Conference on System, Process, and Control, 113-116, Kuala Lumpur, Dec. 2013.
(Summary)
In this study, the objective is to propose a method in which pictures of security cameras are administered automatically. The administered target is abandoned objects. We use a template matching using Genetic Algorithm (GA) for detection of abandoned objects. GA is suitable for global search problems, but it is not suitable for local search problems. Therefore the local domain search technique is included to improve GA property. In GA, a chromosome is composed of the coordinates that locates a template image, and template image's scaling rate and rotation angle. In global domain search, the chromosome has the angle of 0 degree. Thresholds for a fitness function and the number of the generations to start the local domain search are set experimentally. In local domain search, the chromosome's scaling rate and rotation angle of the best individual are changed using a random search method. According to experimental results, detection accuracy is relatively good in the global domain search, but the local domain search is not so effective in some images. In future work, we try to improve the local domain search.
Stephen Githinji Karungaru, Minoru Fukumi and Kenji Terada : Hand Written Character Recognition using Star-Layered Histogram Features, Proceedings of SICE Annual Conference 2013, Dec. 2013.
(Summary)
In this paper, we present a character recognition method using features extracted from a star layered histogram and trained using neural networks. After several image preprocessing steps, the character region is extracted. Its contour is then used to determine the center of gravity (COG). This CoG point is used as the origin to create a histogram using equally spaced lines extending from the CoG to the contour. The first point the line touches the character represents the first layer of the histogram. If the line extension has not reached the region boundary, the next hit represents the second layer of the histogram. This process is repeated until the line touches the boundary of the character's region. After normalization, these features are used to train a neural network. This method achieves an accuracy of about 93% using the MNIST database of handwritten digits.
(Keyword)
文字認識 / image processing / neural network
79.
Momoyo Ito, Kazuhito Sato and Minoru Fukumi : Classification of Driver's Head Posture by using Unsupervised Neural Networks, Proceedings of The Third International Conference on Ambient Computing, Applications, Services and Technologies, 50-57, Porto, Oct. 2013.
(Summary)
We analyze drivers head posture during safety verification at an unsignalized intersection with poor visibility and propose a method for classifying head posture using two types of unsupervised neural networks: Self-Organizing Maps (SOMs) and fuzzy Adaptive Resonance Theory (ART). The proposed method can generate the optimal number of cluster-generated labels for the target problem. We experimentally assess the effectiveness of the proposed method by adjusting the fuzzy ART network vigilance parameters.
80.
Iwase Masashi, Takahashi Keisuke, Minoru Fukumi and Koji Kashihara : Development of an Android Application for Imaging of Superficial Veins, Proc. of SICE Annual Conference 2013, 1515-1517, Nagoya, Sep. 2013.
81.
Momoyo Ito, Kazuhito Sato and Minoru Fukumi : Analysis of Safety Verification Behavior and Classification of Drivers Head Posture, Proceedings of 2013 IEEE International Conference on Mechatronics and Automation, 884-889, Takamatsu, Aug. 2013.
(Summary)
In this paper, we analyze drivers head posture of safety verification at the unsignalized blind intersection, and propose a classification method of head posture using two kinds of unsupervised neural networks: SOMs and Fuzzy ART to quantize drivers head motion for construct a driving assist system which is able to detect the continual deviation signals. The proposed method is able to categorize head posture roughly.
Takako Ikuno, Yohei Takeuchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Abandoned Object Detection by Genetic Algorithm with Local Search, Proceeding of 2013 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'13, 261-264, Kona, Mar. 2013.
(Summary)
In this study, the objective is to propose a method in which pictures of security cameras are administered automatically. The administered target is abandoned objects. We use a genetic template matching method for detection of abandoned objects. In this case, it is necessary to match a whole target image using a template image and it is time-consuming. Therefore we use Genetic Algorithm (GA) with local domain search. GA mainly optimizes the coordinate by which a template image is located, although the size and orientation of the template are fixed for reduction of search time. In addition, those of the template image are charged by the local domain search. In the global domain search using GA, its detection accuracy is relatively good, but the local domain search is not so effective. To solve this problem, we try to improve the local domain search. For this purpose, the template matching of image size and orientation is improved using a deterministic process.
83.
Takashi Fujishima, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method for Detecting Signs of Train Sickness on Tilting Train, Proceeding of 2013 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'13, 341-344, Kona, Mar. 2013.
(Summary)
This paper proposes a method to detect signs of motion sickness based on electroencephalogram (EEG) and heart rate (HR) analysis techniques. In EEG analysis, the averages of power spectra are calculated as the EEG features. In HR analysis, we adopt Lorenz plot. The proposed method discusses differences among human states; normal, sign of motion sickness and motion sickness states, respectively. In order to show the effectives of the proposed method, we conduct experiments using real EEG and HR data.
84.
Takashi Hamano, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Detection of Abandoned Objects Based on Spatiotemporal Features from Public Stationary Camera, Proceeding of 2013 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'13, 257-260, Kona, Mar. 2013.
(Summary)
In this paper, we propose a method of searching abandoned objects based on spatiotemporal features. The final purpose is detection of the abandoned objects from public stationary camera. First, Space-Time (ST) Patch features for limited searching region are computed. These features are essentially composed of 6 features. Our approach is used to efficiently separate moving objects and unmoving ones from background. Next, we detect it as a human or not with HOG features and Real AdaBoost into limited searching regions. The HOG features are converted it one-dimensional histogram of the corresponding values. We learn detection of human with Real AdaBoost by these values.
85.
Momoyo Ito, Kazuhiro Sato, Koichiro Mori and Minoru Fukumi : A Basic Study for Quantification of Driving Behaviors and Estimation of Driving Psychology, Proceeding of 2013 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'13, 21-24, Kona, Mar. 2013.
(Summary)
In this paper, we propose a quantification method of driving behavior using two kinds of unsupervised neural networks: Self-Organizing Maps and Fuzzy Adaptive Resonance Theory, and discuss the relation between safety verification motion and driving psychology. Finally, we show that driving behavior expresses the driving psychology.
86.
Peng Zhang, Yohei Takeuchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Recognition of Eye Motions Using EOG and Statistic Learning, Proceeding of 2013 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'13, 29-32, Kona, Mar. 2013.
(Summary)
Recognition of eye motions has attracted more and more attention of researchers all over the world in recent years. In order to provide an efficient means of communication for patients as ALS (amyotrophic lateral sclerosis) who cannot move even their muscles except eyes. It is important to pursue such a research. In this paper, we propose a new recognition method that uses the LVQ (Learning Vector Quantization) to recognize a class of each motion in the first step, and then uses the relation between max value and min value of each motion's EOG features to separate similar motions in the same class. Using the new method we have obtained a high recognition accuracy of eye motions.
87.
Keisuke Takahashi, Minoru Fukumi and Koji Kashihara : Epibiosis Vein Imaging through Near-Infrared Ray during Temperature Stimuli, Proceeding of 2013 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'13, Kona, Mar. 2013.
88.
Tomo Uchiyama, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Database Optimization Technique for Ethical Pharmaceutical Searching System, Proceeding of 2013 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'13, 145-148, Kona, Mar. 2013.
(Summary)
This paper proposes a method to optimize database for searching ethical pharmaceutical based on multi-objective optimization. The proposed system consists of medicine combination optimization and information visualization to explain intelligible information. The multi-objective genetic programming (MOGP) is used to specify the optimal combinations based on new criteria. By means of computer simulation, the effectiveness of the proposed method is demonstrated.
89.
Yohei Takeuchi, Momoyo Ito and Minoru Fukumi : Nonlinear Eigenspace Models Based on Fast Statistical Learning Algorithm, Proceedings of IASTED International conference on Software Engineering and Applications, 274-278, Las Vegas, Nov. 2012.
(Summary)
In the field of pattern recognition, feature extraction plays an important role prior to classification in order to filter out the background noise, reduce the dimensionality for input and so on. Fisher Linear Discriminant Analysis (FLDA) is well-known as one of the most famous feature extraction methods. In recent years, FLDA has been improved in various ways because an eigenspace is learned faster and/or the classification performance is improved. Simple-FLDA (SFLDA) has been proposed to speed up the learning by improving FLDA algorithm. However, the above methods are calculated in input space. Thus, it might not be efficient in cases where data distribution is complex. Then, Simple Kernel Discriminant Analysis (SKDA), which is an improved version of Kernel Discriminant Analysis (KDA), has been proposed to acquire a better performance for classification by applying kernel trick. Whereas a better performance is acquired by SKDA than that by SFLDA, its learning speed has increased instead. In this paper, an additional improvement is applied to SKDA algorithm and the improved version of SKDA (SIKDA) is introduced. The performance of SIKDA is as same as that of SKDA. In addition, learning speed has become faster than that by SKDA. These are shown in the experiment, especially, the influence of proposed method has seen in a specified dataset.
Momoyo Ito, Kazuhito Sato and Minoru Fukumi : Optimization of Categorizing Driver's Head Motionfor Driving Assistance Systems, Proceedings of SICE Annual Conference 2012 Final Program and Papers, 471-474, Akita, Aug. 2012.
(Summary)
Our system needs quantization of drivers 3D head motions in safety verification only phase variation on 2D image taken by monocular in-vehicle camera, and modeling of head motion information. In this paper, we optimize categorization of drivers head motion using two kinds of unsupervised neural networks: Self-Organizing Maps (SOMs) and Fuzzy Adaptive Resonance Theory (ART).
91.
Yoshimi Miki, Yohei Takeuchi, Momoyo Ito and Minoru Fukumi : Decision of Two Alternatives by EEG using Genetic Algorithm, Proceedings of International Technical Conference on Circuits/Systems, Computers and Communications (CD-R), D-T3-05, Sapporo, Jul. 2012.
(Summary)
The electroencephalograms (EEGs) are a biological signal that represents the electrical activity of the brain. There is a tendency to quantify the human mind and its changes based on preference, feeling, and impression by analyzing electroencephalograms (EEGs) taken on the human scalp. Recently, it is expected that EEG will promote special man-machine interface. In this study we try to estimate human mind and emotion from EEG which becomes popular by development of measurement techniques and analysis of brain activity. In this paper, detection of frequencies, which expressly shows two tendencies to be distinguished while a human subject is in low and high concentration states, is carried out. The purpose of this study is to construct a tool for alternative decision with the frequency difference in two states. Two concentration states can be classified using specified frequencies measured when a human subject is in low and high concentration states. Severe disorders such as Amyotrophic Lateral Sclerosis (ALS) make it hard for motion-impaired people to communicate with others by talking and writing. Therefore, support systems that can help them to establish communication can be constructed with less mental load by utilizing the analyzed data of this study.
92.
Momoyo Ito, Kazuhito Sato and Minoru Fukumi : Classification of Head Motions for Estimation of Driver's Internal States, Proceedings of International Technical Conference on Circuits/Systems, Computers and Communications (CD-R), D-T3-01, Sapporo, Jul. 2012.
93.
Yohei Takeuchi, Momoyo Ito and Minoru Fukumi : Kernel Discriminant Analysis Based on Iterative Calculations, Proceedings of International Technical Conference on Circuits/Systems, Computers and Communications (CD-R), E-T1-04, Sapporo, Jul. 2012.
(Summary)
Fisher Linear Discriminant Analysis (FLDA) is well-known as one of the most famous feature extraction methods in the research field of pattern recognition. In recent years, FLDA has been developed with various techniques to improve its effectiveness for classification, learning speed and generality for any problems. Simple-FLDA (SFLDA), which is a faster version of FLDA, has been proposed to extract features from input datasets in an efficient way. However, it is not always efficient in cases where the datasets are complex, because eigenvectors spanning an eigenspace are acquired in just the input space. In this paper, we proposed a new feature extraction algorithm derived by expanding SFLDA to a non-linear space for effective classification. The algorithm is constructed with just simple calculations, but it is capable of obtaining effective features. In some experiments, better classification accuracies have been acquired with a few specified datasets.
94.
Ryo Yoshioka, Yohei Takeuchi, Momoyo Ito and Minoru Fukumi : Object Search Using a Rough Sketch, Proceedings of International Technical Conference on Circuits/Systems, Computers and Communications (CD-R), E-T1-02, Sapporo, Jul. 2012.
(Summary)
It is difficult to search for lost property in extensive grounds. Therefore, we often have to give up finding it out. To solve this problem, search methods using a security camera have been proposed. It is necessary to not increase the administrative burden when you search an object by using security cameras. In this study, we develop a system for searching an object from camera images automatically. When searching the object, we need to input appropriate information of the object to the system. In this paper, a ``hand-drawn rough sketch'' image is used for this purpose. The present system can identify the object by analyzing the object shape in the image.
95.
Michihiro Jinnai, Neil Boucher, Minoru Fukumi and Taylor Hollis : A new optimization method of the geometric distance in an automatic recognition system for bird vocalisations, Proceedings of the Acoustics 2012 Nantes Conference, 2439-2445, Paris, Apr. 2012.
(Summary)
We have been developing an automatic recognition system for bird vocalisations. Many biologists have been using the early 32 bit version of our system, and we have been working on a 64 bit version. The software segments a waveform of the bird vocalisation from a three-hour continuous recording and extracts the sound spectrum pattern from the waveform using the LPC spectrum analysis. Next, the software compares the sound spectrum pattern (the input pattern) with the standard pattern (that was extracted in advance) using a similarity scale. We use a new similarity scale called the ``Geometric Distance''. The Geometric Distance is more accurate than the conventional similarities in the noisy environment. In the 64 bit version, the software matches an input pattern with the 40,000 elements of the standard patterns per second and per processor, and it is 2.8 times faster than the conventional cosine similarity. In this paper, we introduce an automatic segmentation method of bird vocalisations and a new optimization method of the Geometric Distance. The new optimization method offers improvements of an order of magnitude over the conventional Geometric Distance.
96.
Fumitoshi Taoka, Momoyo Ito, Koji Kashihara and Minoru Fukumi : Detection of Abandoned Objects in Public Facilities, Proceeding of 2012 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'12, 619-622, Honolulu, Mar. 2012.
(Summary)
This paper presents a new method to detect abandoned objects. Our approach consists of the following 5 steps; Dynamic background modeling by using the Gaussian mixture model based on features obtained by space-time patches, human region detection based on Support Vector Machine, movement region detection based on matching of regions histogram features, construction of region database from region features, and abandoned objects detection from the region database. The proposed method adds color information on the features obtained using dynamic background modeling to enhance the effectiveness of detection for an object that is assimilated into background. Experimental results prove the efficiency of our algorithm on PETS2006 and AVSS2007 benchmark data.
97.
Kazuya Yaegashi, Momoyo Ito, Koichiro Mori, Kazuhito Sato, Koji Kashihara and Minoru Fukumi : Fundamental Study on EEG Analysis for Safety Driving Support System, Proceeding of 2012 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'12, 587-590, Honolulu, Mar. 2012.
(Summary)
In this paper, as a fundamental study of relationship between drivers internal state and the safety verification motion, we analyze four drivers EEG (electroencephalogram) at unsignalized intersections, and discuss effects of drivers internal state changes on safety verification.
98.
Takahide Funabashi, Yohei Takeuchi, Momoyo Ito, Koji Kashihara and Minoru Fukumi : Recognition of FingerMotion by Wrist EMG, Proceeding of 2012 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'12, 433-436, Honolulu, Mar. 2012.
(Summary)
In this paper finger motions were recognized by EMG measured using dry-type electrodes attached to wrist. Target behaviors to be recognized are four motions that the Janken ``rock'', ``scissors'', ``paper'' and when not doing anything ``neutral''. We tried to reduce an execution time by the simple-PCA in training and recognition, with a view to implementation of interface which can be utilized in daily life.
99.
Yuki Ikami, Koji Kashihara, Momoyo Ito and Minoru Fukumi : Visual Illusion of Depth Percception during Car Driving, Proceeding of 2012 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'12, 293-296, Honolulu, Mar. 2012.
100.
Yusuke Yamamura, Yohei Takeuchi, Momoyo Ito, Koji Kashihara and Minoru Fukumi : Classification of Motions by EMG of Ankle, Proceeding of 2012 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'12, 285-288, Honolulu, Mar. 2012.
101.
Natsumi Ohtani, Yohei Takeuchi, Momoyo Ito, Koji Kashihara and Minoru Fukumi : Speech Recognition of Whisper Voice Based on EMG Signals, Proceeding of 2012 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'12, 281-284, Honolulu, Mar. 2012.
(Summary)
Speech recognition using surface electromyographic (EMG) is regarded as one of substitute or support for situations that cannot obtain a clear sound. This research aims at Japanese speech recognition to a whisper voice using EMG signals. In this paper, as the first step, we recognized Japanese 5 vowels of the prolonged sound. In order to recognize them, FFT was carried out for the EMG signals obtained from 2 muscular parts. After that, the features were extracted by Simple-PCA, and 1-NN was used to recognize them. The average recognition accuracy of 5 vowels was about 79%.
102.
Stephen Githinji Karungaru, Kenji Terada and Minoru Fukumi : Feature Extraction For Human Recognition Based On Bifurcation Points Using Genetic Algorithms, Proc. of NCSP'2012, 615-618, Honolulu, Mar. 2012.
Kentaro Mori, Satoru Tsuge, Shingo Kuroiwa, Momoyo Ito, Koji Kashihara and Minoru Fukumi : Acoustic Model Selection Method for Speaker Dependent Speech Recognition, Proceeding of 2012 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'12, 158-161, Honolulu, Mar. 2012.
(Summary)
In this paper, we investigate an intra-speaker's variability using a specific speaker's speech data witch are collected over long time period. Especially, we investigate a relationship between a speech recognition performance and a distance calculated by dynamic programming matching algorithm (we called ''DP distance'' hereafter), and a time dif- ference of a speech recognition day and an acoustic model training day. From the investigation results, we can see that the intra-speaker's speech variability varies the recognition performance. Hence, for restricting the variation of speech recognition performances caused by intra-speaker's speech variability, we proposed an acoustic model selection method in this paper. The proposed method decides candidates of optimal acoustic model by using DP distance, and selects an optimal acoustic model by using likelihood. For evaluating the proposed method, we conduct speech recognition experiment using a male speaker's speech data collected over long time period. As a result, speech recognition accuracy of selected model by the proposed method is slightly degraded than that of the conventional method. However, the proposed method can select an optimal acoustic model with smaller calculation costs than the conventional method.
104.
Den Nagarekawa, Koji Kashihara, Momoyo Ito and Minoru Fukumi : A Cloth Simulation System to Select the Right SIze, Proceeding of 2012 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'12, 49-51, Honolulu, Mar. 2012.
105.
Yohei Takeuchi, Momoyo Ito, Koji Kashihara and Minoru Fukumi : A Novel Nonlinear Discriminant Analysis by Iterative Operations, Proceeding of 2012 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'12, 104-107, Honolulu, Mar. 2012.
(Summary)
In the past, Principal Component Analysis (PCA), Fisher Linear Discriminant Analysis (FLDA) and other feature analyses have been widely used as statistical techniques in pattern recognition. Furthermore, these techniques have been improved to extract better features for both classification and learning. For instance, Simple-PCA (SPCA) and Simple-FLDA (SFLDA) have been proposed to learn an eigenspace not only easier but also faster by iterative operations. However, the eigenspace by these feature extraction algorithms might not be adequate to every dataset, because these are linear feature extraction methods. In this paper, we propose a new Simple-FLDA algorithm applied Kernel method, which leads nonlinear feature extraction. Therefore, the proposed algorithm is simply constructed. In addition, kernel base vectors, which are projected from high-dimensional space to inner product space, are used for feature extraction. The same criterion for discriminant analysis is used by using these kernel base vectors. In the experiment, 10% better classification accuracy is acquired when one dataset has high dimensionality.
106.
Stephen Githinji Karungaru, Kenji Terada and Minoru Fukumi : Abandoned Luggage Detection using a Dynamic Background Model and Earth Movers Distance, Proceedings of 18th Korea-Japan Joint Workshop on Frontiers of Computer Vision, 82-88, Yokohama, Japan, Feb. 2012.
107.
Koji Kashihara, Momoyo Ito and Minoru Fukumi : Estimation of venous shapes acquired from CMOS camera images., Proceedings of the Eighteenth Korea-Japan Joint Workshop on Frontiers of Computer Vision (FCV2012), 47-52, Kawasaki, Feb. 2012.
108.
Momoyo Ito, Kazuhito Sato, Minoru Fukumi and Ikuro Namura : Brain Tissues Segmentation for Diagnosis of Alzheimer-Type Dementia, Proceedings on IEEE Nuclear Science Symposium, Medical Imaging Conference, 3847-3849, Valencia, Oct. 2011.
(Summary)
In this paper, we specifically discuss segmentation of brain tissues which are used for calculation of atrophy rate. We proposed a brain tissue segmentation method using two kinds of unsupervised neural networks: Self-Organizing Maps (SOMs) and Fuzzy Adaptive Resonance Theory (ART). The performance of proposed method was tested in two brain MR images used in daily diagnosis. Proposed method could segment CSF accurately with continuity of brain tissues.
Momoyo Ito, Kazuhito Sato, Hirokazu Madokoro, Koji Kashihara and Minoru Fukumi : Basic Studies for Estimation of Driver's Internal States Using Head Positions, Proceedings on 4th International Symposium on Applied Sciences In Biomedical and Communication Technologies (ISABEL), Barcelona, Oct. 2011.
(Summary)
In this paper, we analyze driving movies taken by monocular in-vehicle camera, and examine drivers head position category in safety verification at intersections for quantification of head motion information. Moreover, we propose a quantifiable categorizing algorithm of head motion using two kinds of unsupervised neural networks. Through an experiment on actual driving data, the results provide a possibility of quantification of individual head position in safety verifications.
110.
Koji Kashihara, Momoyo Ito and Minoru Fukumi : Development of automatic filtering system for individually unpleasant data detected by pupil-size change., Proceedings of 2011 IEEE International Conference on Systems, Man, and Cybernetics, 3311-3316, Anchorage, Oct. 2011.
Koji Kashihara, Momoyo Ito and Minoru Fukumi : An analytical method for face detection based on image patterns of EEG signals in the time-frequency domain., Workshop on Brain-Machine Interfaces, Proceedings of 2011 IEEE International Conference on Systems, Man, and Cybernetics, 25-29, Oct. 2011.
Koichirou Mori, Momoyo Ito, Kazuhito Sato, Koji Kashihara and Minoru Fukumi : Analysis of Relationship between Head Motion Information and Driving Scene for Dangerous Driving Forecast, Proc. of SICE Annual Conference 2011, 2705-2709, Tokyo, Sep. 2011.
(Summary)
In this study, we are aiming to detect deviation signal which has a possibility to cause car accidents, using drivers head motion information. The head motion information is extracted from only 2D camera images. Our method is able to extract the information without alignment of facial positions. In this paper, SOMs categorization results of our method are discussed and evaluated.
113.
Yohei Takeuchi, Momoyo Ito, Koji Kashihara and Minoru Fukumi : Supervised Iterative Learning Algorithm for Eigenspace Models, Proc. of SICE Annual Conference 2011, 2361-2365, Tokyo, Sep. 2011.
(Summary)
In pattern recognition, the principal component analysis (PCA) is one of the most famous feature extraction methods for dimensionality reduction of high-dimensional datasets. Furthermore, Simple-PCA (SPCA) which is a faster version of the PCA, has been carried out effectively by iterative operated learning. However, in SPCA, when input data are distributed in a complex way, SPCA might not be efficient because it is learned without class information of the dataset. Thus, SPCA cannot be said that it is optimal for classification. In this paper, we propose a new learning algorithm, which is learned with the class information of the dataset. Eigenvectors spanning eigenspace of the dataset are obtained by calculation of data variations belonging to each class. We will show the derivation of the proposed algorithm and demonstrate some experiments to compare the SPCA with the proposed algorithm by using UCI datasets.
Tadahiro Oyama, Higashi Keita, Choge Hillary, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : BCI Based on Single EEG Used Simple Electroencephalograph, Proc. of SICE Annual COnference 2011, 2331-2334, Tokyo, Sep. 2011.
115.
Masato Miyoshi, Mori Kentaro, Kashihara Yasunori, Nakao Masafumi, Satoru Tsuge and Minoru Fukumi : Personal Identification Method using Footsteps, Proc. of SICE Annual COnference 2011, 1615-1620, Tokyo, Sep. 2011.
(Summary)
Some researches began to study the personal authentication using the footsteps which are used as the biological information. In general, the footstep identification methods are consist of a feature extraction method and a classification method. In the feature extraction method, the useful features, which are a power spectrum, a melcepstrum, a gait cycle, and so on, are extracted from audio signals of the footsteps. Using these features, the personal identification are performed by using Dynamic TimeWarping (DTW), k-means clustering method, and Support Vector Machine (SVM) in the classification method. However, it has not been established the suitable methods for the personal identification using the footsteps. Hence, we investigate the feature extraction method and the classification method. In this paper, we propose an identification method which uses Mel-Frequency Cepstrum Coefficients (MFCCs) as footstep features, k-Nearest Neighbor (k-NN) which uses DTW as a distance measure and Gaussian Mixture Models (GMMs) as classifiers.
Shin-ichi Ito, Masashi Hamaguchi, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : Variability in EEG with Single Point Sensing as Inter-Individual Difference Measure Using Self-Organizing Map, Proc. of 2011 International Symposium on Nolinear Theory and its Applications, NOLTA2011, 290-293, Kobe, Sep. 2011.
(Summary)
In this paper, we introduce an EEG analysis technique to confirm an inter-individual difference in prefrontal cortex EEG with a single point sensing. The device for recording the EEG uses the dry-type sensor and a few numbers of electrodes. The EEG analysis adapts the feature mining on EEG pattern using a self-organizing map (SOM). The EEG patterns are determined based on the preference evaluation on sound listened to. In the preprocessing, we extract the EEG feature vector by calculating the time average on each frequency band which are theta, low-alpha, high-alpha, low-beta, high-beta, respectively. To confirm the inter-individual difference, we do experiments using real EEG data. These results show that the learning results by SOM on each human are clearly different when using same initial weight values for the SOM.
117.
Yohei Takeuchi, Momoyo Ito, Koji Kashihara and Minoru Fukumi : Novel Supervised Feature Extraction Algorithm Based on Iterative Calculations, Proc. of The IEEE International Conference on Information Reuse and Integration (IRI2011), 304-308, Las Vegas, Aug. 2011.
(Summary)
In pattern recognition, the principal component analysis (PCA) is one of the most famous feature extraction methods for dimensionality reduction of high-dimensional datasets. Furthermore, Simple-PCA (SPCA) which is a faster version of the PCA, has been carried out effectively by iterative operated learning. However, in SPCA, when input data are distributed in a complex way, SPCA might not be efficient because it is learned without class information of the dataset. Thus, SPCA cannot be said that it is optimal for classification. In this paper, we propose a new learning algorithm, which is learned with the class information of the dataset. Eigenvectors spanning eigenspace of the dataset are obtained by calculation of data variations belonging to each class. We will show the derivation of the proposed algorithm and demonstrate some experiments to compare the SPCA with the proposed algorithm by using UCI datasets.
Yohei Takeuchi, Momoyo Ito, Koji Kashihara and Minoru Fukumi : Supervised Feature Extraction Algorithm by Iterative Calculations, Proc. of The 2nd Conference on Next Generation Information Technology (ICNIT2011), 46-49, Gyounju, Jun. 2011.
(Summary)
In pattern recognition, the principal component analysis (PCA) is one of the most famous feature extraction methods for dimensionality reduction of high-dimensional datasets. Furthermore, Simple-PCA (SPCA) which is a faster version of the PCA, has been carried out effectively by iterative operated learning. However, in SPCA, when input data are distributed in a complex way, SPCA might not be efficient because it is learned without class information of the dataset. Thus, SPCA cannot be said that it is optimal for classification. In this paper, we propose a new learning algorithm, which is learned with the class information of the dataset. Eigenvectors spanning eigenspace of the dataset are obtained by calculation of data variations belonging to each class. We will show the derivation of the proposed algorithm and demonstrate some experiments to compare the SPCA with the proposed algorithm by using UCI datasets.
119.
Masato Miyoshi, Satoru Tsuge, Tadahiro Oyama, Momoyo Ito and Minoru Fukumi : Feature Selection Method for Music Mood Score Detection, Proc. of ICMSAO'2011, 713-718, Kuala Lumpur, Apr. 2011.
(Summary)
In general, music retrieval and classification methods using music moods use a lot of acoustic features similar to music genre classification. These features are used as the spectral features, the rhythm features, the harmony features, and so on. In this paper, we propose a feature selection method for detecting music mood scores. In the proposed method, features which have strong correlation with mood scores are selected from a lot of features. Then, these are input into Multi-Layer Neural Networks (MLNNs) and mood scores are detected every mood labels. For evaluating the proposed method, we conducted the music mood score detection experiments. Experimental results show that the proposed method improves the detection performance compared to not use the feature selection.
Takuya Shiraishi, Atsushi Ishitani, Momoyo Ito, Stephen Githinji Karungaru and Minoru Fukumi : Operation Improvement of Indoor Robot by Gesture Recognition, Proc. of ICMSAO'2011, 572-575, Kuala Lumpur, Apr. 2011.
(Summary)
Recently, the demand for the indoor robots has increased. Therefore, increased opportunities for many people to operate the robots have emerged. However, for many people, it is often difficult to operate a robot using the conventional methods like remote control. To solve this problem, we propose a robot operation system using the hand gesture recognition. Our method pays attention to the direction and movement of the hand. We were able to recognize several gestures in real-time.
Stephen Githinji Karungaru, Kenji Terada and Minoru Fukumi : Background Updating for Abandoned Luggage Detection at Train Stations, Proc. of ICMLC'2011, Vol.2, 11-14, Singapore, Feb. 2011.
122.
Yohei Takeuchi, Momoyo Ito and Minoru Fukumi : Novel Approximate Stastical Learning Algorithm for Large Complex Datasets, Proc. of ICMLC'2011, Vol.3, 236-239, Singapore, Feb. 2011.
(Summary)
In pattern recognition, simple Principal Component Analysis (SPCA) method is one of the most famous feature extraction methods for dimensionality reduction of high-dimensional image datasets. However, when input data are distributed in a complex way, SPCA which is a linearly feature extraction method, might not be efficient. In such cases, kernel method, which is nonlinear feature extraction method, was shown its high generalization capability. In this paper, we apply kernel method to SPCA, and show the derivation. Finally, it was tested and compared with original SPCA algorithm by using Japanese paper currency images.
123.
Atsushi Ishitani, Takuya Shiraishi, Stephen Githinji Karungaru, Momoyo Ito and Minoru Fukumi : A Simple Interface for Mobile Robot Using Motion Stereo Vision, Proc. of ICMLC'2011, Vol.2, 174-178, Singapore, Feb. 2011.
(Summary)
In remote control, user interfaces are very important and it is desired that they have intuitive operability and be simple systems. However, presently, most of interfaces have not satisfied these needs. Therefore, we propose a simple interface to control mobile robots equipped with a camera by instructing it to move to a desired location using the image set to a screen. For the robot to control its movement, it needs to calculate the three-dimensional information of the instructed point. In this work, we calculate the three-dimensional position using motion stereo vision, and perform experiments using a wheeled mobile robot with a single camera in a real environment.
124.
Hitoshi Takano, Stephen Githinji Karungaru, Momoyo Ito and Minoru Fukumi : Stop Sign Recognition from Drive Scenes, Proc. of ICMLC'2011, Vol.1, 549-552, Singapore, Feb. 2011.
125.
Stephen Githinji Karungaru, Kenji Terada and Minoru Fukumi : Improving mobility for blind persons using video sunglasses, Proceedings of 17th Korea-Japan Joint Workshop on Frontiers of Computer Vision, 1-5, Ulsan, South Korea, Feb. 2011.
(Summary)
Blind people navigate safely through a familiar room based on strong expectations about the location of objects. If something has been moved, added or removed, it can present a difficulty and potentially a danger. Human eyes are one of the most important body parts that help humans to understand and interact with their surroundings. Most learning and recognition of objects around us is accomplished using the eyes. Given the recent advancement of imagery systems and ever increasing processing power of microprocessors, a machine vision aiding system for the blind can be a reality. In the initial system we propose a system consisting of camera equipped sunglasses to capture the images and pattern recognition to automatically detect Braille tiles to aid mobility of blind persons. The information obtained is passed to the subject using audio messages.
Hironobu Fukai, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : Robust Age Estimation for Face Posture Change, the 8th edition of France-Japan and 6th Europe-Asia Congress on Mechatronic 2010, 508-513, Yokohama, Nov. 2010.
127.
Momoyo Ito, Kazuhito Sato, Ikuro Namura and Minoru Fukumi : Extraction of Brain Regions for Image Diagnosis of Alzheimer-type Dementia Based on Atrophy Progress Speeds, Conference Record of 2010 IEEE Nuclear Science Symposium and Medical Imaging Conference, M19-375, Knoxville, Tennessee, USA, Nov. 2010.
(Summary)
In this paper, we propose an extraction method of the ROI fitting to individual brain shape and size using two deformable models hierarchically. Moreover, we discuss diagnostic application using atrophy progress speeds briefly.
Hironori Takimoto, Akira Yoshida, Yasue Mitsukura and Minoru Fukumi : Invisible Print-type Calibration Pattern based on Human Visual Perception, Proceeding of Int. Conf. on Image Processing'2010, 2601-2604, Hong Kong, Sep. 2010.
(Summary)
In the print-type steganographic system and watermark, a calibration pattern is arranged around contents where invisible data is embedded, as plural feature points between an original image and the scanned image for normalization of the scanned image. However, it is clear that conventional methods interfere with page layout and artwork of contents. In addition, visible calibration patterns are not suitable for security service. In this paper, we propose an arrangement and detection method of an invisible calibration pattern based on human visual perception. We embed the calibration pattern in an original image by adding high frequency component to blue intensity in a limited region. Moreover, the proposed calibration pattern protects page layout and artwork.
129.
Yoshiki Kubota, Takuya Akashi, Minoru Fukumi and Yoshisuke Kurozumi : Investigation of Accuracy for drawing of Object Detection Method with Sketch, Proc. of World Automation Congress (WAC'2010), 1-6, Kobe, Sep. 2010.
(Summary)
There are many image retrieval systems. In these systems, it is difficult to search an object which can not describe in words. Moreover, tagging with words and prior training are necessary. To solve these problems, a purpose of our study is a intuitive and visual image retrieval system using a sketch. As a basis of the system, object detection method with sketch have been proposed. In this paper, an effect of the difference of various sketches in the proposed method is investigated by an experiment. As a results, it is shown that the proposed method is influenced by a detailed part of the sketch, and too sensitive detection is achieved.
130.
Maiko Kudo, Takuya Akashi, Minoru Fukumi and Yoshisuke Kurozumi : Improvement of Likelyhood in Particle Filter for Interactive Color Tracking, Proc. of WAC'2010, 1-6, Kobe, Sep. 2010.
(Summary)
In this paper, we propose an idea that is improving method of likelihood in particle filter. Following methods have been proposed as the previous method by us. First, an user is able to select a tracking object as a region of interest (ROI) in a single image from video sequence. Then, color feature in the ROI are used as a color of interest (COI). During the tracking, two color components of HSV color space are automatically chosen by the color feature analysis in each particle. However, in the previous method, values of likelihood is fixed. If values of particles are in range of the COI, the likelihood value is 1.0. Otherwise particles are set in 0.2 as the likelihood value. This value is fixed regardless of the distance from COI. To improve accuracy of tracking, we propose the method that the likelihood is dynamically changed depending on the distance from COI. In our experiments, the proposed method is compared with the previous method. As results, the accuracy of the proposed method is better than the previous method.
131.
Hironori Takimoto, Seiki Yoshimori, Yasue Mitsukura and Minoru Fukumi : Classification of Hand Postures Based on 3D Vision Model for Human-Robot Interaction, Proc. of 19th IEEE International Symposium on Robot and Human Interactive Communication (Ro-Man'10), 312-317, Viareggio, Sep. 2010.
(Summary)
In this paper, a method for hand posture recognition, which is robust for hand posture changing in an actual environment, is proposed. Conventionally, a data glove device and a 3D scanner have been used for the feature extraction of hand shape. However, the performance of each approach is affected by hand posture changing. Therefore, this paper proposes the posture fluctuation model for efficient hand posture recognition, based on 3D hand shape and color feature obtained from a stereo camera. A large set of dictionary for posture recognition is built by various leaned hand images which were auto-created from one scanned hand image, based on plural proposed models. In order to show the effectiveness of proposed method, performance and processing times for posture recognition are compared to conventional method. In addition, we perform the evaluation experiment by using the Japanese sign language.
Shin-ichi Ito, Yasue Mitsukura, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : Association between Ego Scores and Individual Characteristics in EEG Analysis, --- Basic Study on Individual Brain Activity ---, Proc. of 19th IEEE International Symposium on Robot and Human Interactive Communication (Ro-Man'10), 230-235, Viareggio, Sep. 2010.
(Summary)
This paper introduces a method for discussing an association between ego scores and individual characteristics in EEG analysis. The ego scores express one of the personality and/or nature on a human. The EEG analysis calculates the power spectra of the frequency weighted of the EEG signal, and evaluates through detecting the EEG patterns while listening to the music. A k-nearest neighbor classifier is used to detect the EEG patterns. The specified weight value to weight frequency of EEG and the results of EEG pattern detection express the individual characteristics in EEG analysis. Finally, we do experiment using a real EEG data for discussing the association the ego scores and the individual characteristics in EEG analysis. An interesting tendency was that the subjects who were the combined ego-type tended to have a powerful response to negative stimuli more than positive stimuli. They had the stable filter weighted frequency.
Masahito Miyoshi, Hillary Kipsang Choge, Satoru Tsuge, Tadahiro Oyama, Momoyo Ito and Minoru Fukumi : Music Impression Detection Method for User Independent Music Retrieval System, Proc. of KES'2010, 612-621, Wales (U.K.), Sep. 2010.
(Summary)
In previous work, we have proposed the automatic sensitive word score detection system for a user dependent music retrieval system. However, the user dependent method causes a lot of burdens to the user because the system requires a lot of data for adapting it to each user. Hence, in this paper, we propose an automatic sensitive word score detection method for a user independent music retrieval system and evaluate the proposed method using 225 music data. Experimental results show that 87.5% of music patterns succeeded in detection of sensitive word score in the case that the difference between estimated and evaluated score is 1 (Error 1 rate). Moreover, we conduct subjective evaluation experiments to evaluate the proposed method as a utility method. From this experiment, it is observed that the user satisfaction level of the proposed method is higher than random selection impression detection.
134.
Hironobu Fukai, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : Age Estimaton Using Kernel Regression Analysis, Proceeding of International Symposium on Nonlinear Theory and its Applications (NOLTA2010), 221-224, Krakow, Poland, Sep. 2010.
135.
Hironori Takimoto, Seiki Yoshimori, Yasue Mitsukura and Minoru Fukumi : An Embedding and Detection Method of Invisible Calibration Pattern for Print-Type Data Hiding, Proceeding of International Symposium on Nonlinear Theory and its Applications (NOLTA2010), 217-220, Krakow, Poland, Sep. 2010.
136.
Hironori Takimoto, Seiki Yoshimori, Yasue Mitsukura and Minoru Fukumi : Invisible Calibration Pattern based on Human Visual Perception Characteristics, Proceeding of Int. Conf. on Pattern Recognition'2010, 4210-4213, Istanbul, Aug. 2010.
(Summary)
In the print-type steganographic system and watermark, a calibration pattern is arranged around contents where invisible data is embedded, as plural feature points corresponding to between an original image and the scanned image for normalization of the scanned image. However, it is clear that conventional methods interfere with page layout and artwork of contents. In addition, visible calibration patterns are not suitable for security service. In this paper, we propose an arrangement and detection method of an invisible calibration pattern based on characteristics of human visual perception. The calibration pattern is embedded to blue intensity in an original image by adding high frequency component.
137.
Michihiro Jinnai, Yukio Akashi, Shinsuke Nakaya, Fuji Ren and Minoru Fukumi : Recognition of Abnormal Vibrational Responses of Signposts using the Two-dimensional Geometric Distance and Wilcoxon Test, Proc. IEEE International Conference on Natural Language Processing and Knowledge Engineering, 166-173, Beijing, Aug. 2010.
(Summary)
In expressway companies, workers have been impacting signposts using wooden hammers and estimating the degree of the corrosion by listening to the sound. In order to automate this, we have been developing software that recognizes an abnormal impact vibrational response due to corrosion. This software extracts sonograms from impact vibrational waves using the LPC spectrum analysis, and matches images of the sonogram between a standard and an input impact vibrations using the Two-dimensional Geometric Distance. Then, the software distinguishes the abnormality of the input impact vibration using Wilcoxon rank-sum test. We have measured the impact vibrations of five normal signposts and five abnormal signposts, and carried out the automatic recognition experiments. As a result, the software has recognized correctly in all cases. We have verified the effectiveness of the proposed method.
Shin-ichi Ito, Yasue Mitsukura, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : Study on Association between User's Personality and Individual Characteristic of Left Prefrontal Pole EEG Activity, Proc. of 2010 Sixth International Conference on Natural Computation (ICNC'10), 2163-2166, Yantai, Aug. 2010.
(Summary)
This paper introduces a method for discussing the association between human personality based on egogram scores and the results of classifying the electroencephalogram (EEG) patterns while listening to the music. The egogram based on psychological testing is used for analyzing his/her personality. The frequencies of the EEG analyzed are the components that contain significant and immaterial information and have different importance. We express the different importance through the weight value on frequencies using real-coded genetic algorithm. Then, the EEG patterns, which are determined based on the evaluation of the impression on the music, are classified using the k-nearest neighbor method. Finally, we discuss the association between his/her personality and the individual characteristic in the EEG analysis. An interesting tendency was that the false-detection accuracy of the EEG pattern, meaning the response on negative stimuli, did not become low when the score and level on ego states of Adult was extreme.
139.
Shin-ichi Ito, Yasue Mitsukura, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : Study on Relationship between Personality and Individual Characteristic of EEG for Personalized BCI, Proc. of IEEE Region 8 SIBIRCON-2010, 106-111, Irkutsuk, Jul. 2010.
(Summary)
It is an important issue that the relationship between personality and individual characteristic of an electroencephalogram (EEG) for using a personalized brain computer interface on daily basis. In this paper, we introduce a method for discussing the relationship between personality based on egogram scores and the results of the EEG pattern detection. The egogram based on psychological testing is used for analyzing human personality. The frequencies of the EEG analyzed are the components that contain significant and immaterial information and have different importance. We express the different importance through the weight value on frequencies using real-coded genetic algorithm. Then, the EEG patterns, which are determined based on the evaluation of the impression on the music, are classified/detected using the k-nearest neighbor method. Finally, we discuss the relationship between human personality and the individual characteristic in the EEG analysis. An interesting tendency was that the false-detection ratio of the EEG pattern, meaning the response on negative stimuli, did not become low when the score and level on ego states of Adult was extreme.
Hiroyuki Fukai, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : Age and Gender Estimation by using Facial Image, Proceeding of AMC'2010, ND-003301_1-ND-003301_4, Nagaoka, Mar. 2010.
(Summary)
In this paper, we propose age and gender estimation system by vairous features. Age and gender has a lot of characteristics. These characteristics are one of the difficult cognitive process in human interaction. If we can extract the important feature of this cognitive process, it is considered that the age and gender estimation by the machine becomes possible. Therefore, we propose a method of age and gender feature extraction and estimation using the face image. In this paper, the age of the face means apparent-age that is based on the human perception of age. Moreover, person's aging and gender difference appear in the faces. For example, the pigmented spot, the wrinkle, sagging skin, shape, color of skin, and so on. Thus, we extract these several features for age and gender estimation. Furthermore, we estimate a continuous age and gender using a neural network (NN).
141.
Seiki Yoshimori, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : A Proposal of Feature Extraction for Impression Analysis, Proceeding of AMC'2010, ND-000957_1-ND-000957_4, Nagaoka, Mar. 2010.
(Summary)
The concern of image processing has increased, in recent year. However, the evaluation of facial impression that we are naturally doing from ones face in our daily life is not treated in this area. The automatic facial impression evaluation applies for wide area. Then we analyzed important facial featres for impression analysis, in order to achieve automatic evaluation of facial impression. In order to evaluate the proposed method, we performed computer simulation. As the result of this simulation, we confirmed that important facial features for facial impression are difference while gender and age.
142.
Hironori Takimoto, Seiki Yoshimori, Yasue Mitsukura and Minoru Fukumi : Invisible Calibration Pattern Based on Human Visual Perception, Proceeding of AMC'2010, 159-163, Nagaoka, Mar. 2010.
(Summary)
In the print-type steganographic system and watermark, a calibration pattern is arranged around contents where invisible data is embedded, as plural feature points between an original image and the scanned image for normalization of the scanned image. However, it is clear that conventional methods interfere with page layout and artwork of contents. In addition, visible calibration patterns are not suitable for security service. In this paper, we propose an arrangement and detection method of an invisible calibration pattern based on human visual perception. We embed the calibration pattern in an original image by adding high frequency component to blue intensity in a limited region. Moreover, the proposed calibration pattern protects page layout and artwork.
143.
Toru Seo, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Driver Alcohol Consumption Detection System using EEG, Proceeding of 2010 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'10, 632-635, Honolulu, Mar. 2010.
144.
Kentaro Nakao, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Biped Robot Walking on Uneven Surfaces using Reinforcement Learning, Proceeding of 2010 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'10, 572-575, Honolulu, Mar. 2010.
(Summary)
Recently, reinforcement learning is one of the approaches focused on to achieve adaptive motion in unknown environment. However, most existing approaches require a large number of trials, which becomes one the major problems when the method is applied to an actual robot. Therefore, in this study, we propose an approach that attains low calculation cost to adapt the robot to unknown environments using reinforcement learning. To increase learning accuracy and efficiency, three patterns of rewards are tried and compared for verification. As a result, 70% success rate walking on small uneven surface was achieved
145.
Atsushi Ishitani, Shiraishi Takuya, Fumitoshi Taoka, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : A Simple Interface for Mobile Robot using Motion Stereo Vision, Proceeding of 2010 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'10, 560-563, Honolulu, Mar. 2010.
(Summary)
Recently, various kinds of robots are present close to our living environment from their original limited usage in factories, etc. Especially, remote-controlled robots with mobile actuator are being put to practical use for some purposes such as conveying systems, home and office security etc, because the use of remote-controlled robots is easier than autonomous robots. These mobile robots require a controller. In this work, we develop a simple interface for mobile robots. To uses this interface, the users need only point to the position on a screen where they want the robot to move to. After that, the robot controls its own movement between the current position and the target position. To autonomously move to the selected point, the robot needs to calculate the direction and distance of the target from its own position. To calculate direction, we use a simple camera model and pixels surrounding the target point to extract a template. Template Matching is then used to recognize the point during movement. Our interface can forecast movement zone of template image, because it controls movement of the robot. To calculate distance, we use Motion Stereo Vision (MSV). We have two reasons for using this method. First, MSV needs only one camera therefore cost is reduced, and secondly, MSV involves camera movement which is well suited for the mobile robot we are constructing.
146.
Hironori Takimoto, Seiki Yoshimori, Yasue Mitsukura and Minoru Fukumi : Hand Gesture Recognition under Variable Pose Based on 3-D Appearance Model, Proceeding of 2010 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'10, 504-507, Honolulu, Mar. 2010.
147.
Seiki Yoshimori, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : Change of the important impression part in the face according to the age, Proceeding of 2010 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'10, 445-448, Honolulu, Mar. 2010.
(Summary)
The achievement of the automatic impression evaluation can apply to wide area application such as automatic generation of CG, automatic likeness making in a criminal investigation, cosmic sergeries, and entertainments. However the impression evaluation that we are naturally doing in daily life is not treated conventional researches. Then we stuy the relationship of the change between facial impression and facial part by age in this paper.
148.
Masafumi Nakao, Satoru Tsuge, Minoru Fukumi and Shingo Kuroiwa : Speaker vector combination method of air- and bone-condction speech for speaker identification, Proceeding of 2010 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'10, 417-420, Honolulu, Mar. 2010.
(Summary)
Recently, some new sensors, such as bone-conductive microphones, throat microphones, and non-audible (NAM) microphones, besides conventional condenser microphones have been developed for collecting speech data. We focus on boneconduction speech data collected by the bone-conductive microphone. Generally, a Gaussian Mixture Models (GMMs) are used for the speaker recognition. However, recognition performance of these models is low under the condition that the number of enrollment data is small. Hence, the speaker recognition method using speaker vectors has been proposed. It is reported that this method is useful for the speaker recognition when the number of enrollment data is small. Accordingly, in this paper, we propose a speaker vector combination method of using air- and bone- conduction speech for the speaker identification. The proposed method individually trains the speaker vector of air-conduction speech and the speaker vector of bone-conduction speech. After that the proposed method combines these speaker vectors and uses these for speaker identification. For evaluating the proposed method, we conducted 100 female speakers speaker identification experiment. Experiment results show that the speaker identification accuracy of the proposed method is able to improve that of each speech. In fact, 60.1% of the accuracy of the proposed method has improved 53.3% of air-conduction speech.
149.
Miyoshi Masato, Tadahiro Oyama, Hillary Choge, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Music Classification using Sensitive Words, Proceeding of 2010 International Workshop on Nonlinear Circuits, Communication and Signal Processing NCSP'10, 211-214, Honolulu, Mar. 2010.
(Summary)
In this paper, we propose an automatic sensitive word score detection method for a music retrieval system. The proposed method consists of two parts, which are a feature extraction part and a score detection part. The score detection part detects a sensitive word score of each music. In this paper, we use an impression represented by 7 levels for the sensitive word score and the Multi-Layer Neural Networks (MLNNs) are used for detector of sensitive word score. Moreover, we investigate the effective features to detect the sensitive word score. To evaluate the proposed method, we conduct a sensitive word score detection experiment.
150.
Kazuhiro Nakaura, Tadahiro Oyama, Satoru Tsuge, Takuya Akashi and Minoru Fukumi : Improved Feature Generation Ppoperty in Fast Statistical Learning Algorithm, Proc. of IWAIT'2010, No.Paper-121, 1-5, Kuala Lumpur, Jan. 2010.
(Summary)
This paper presents a new accuracy improvement technique in the statistical learning algorithm Simple-FLDA (Simple-Fisher Linear Discriminant Analysis) which was presented by authors. Recently, iterative learning algorithms for obtaining eigenvectors in the principal component analysis (PCA) have been presented in the field of pattern recognition and neural network. On the one hand, the Fisher linear discriminant analysis (FLDA) has been used in many fields, including especially face image analysis, and compared to PCA in relation to feature generation. However there are drawbacks of FLDA such as a long computational time based on a large-sized covariance matrix, the limit of the number of effective eigenvectors, and the issue that the within-class covariance matrix is usually singular. In order to overcome this difficulty, we proposed the feature generation method Simple-FLDA which is approximately derived from geometrical interpretation of FLDA. This algorithm is similar to the Simple-PCA and does not need matrix operation. In this paper, a new calculation technique of eigenvectors in the Simple-FLDA is presented. Its preliminary simulation results are given for a simple face recognition problem and others.
151.
Fumitoshi Taoka, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Detection of Abandoned Objects from Fluctuating Environment Using Active Background Subtraction, Proc. of IWAIT'2010, No.paper-141, 1-4, Kuala Lumpur, Jan. 2010.
(Summary)
This paper proposes a new method to detect abandoned objects with a focus on Active Background Subtraction. Some estimate background models for dealing with fluctuating environment are utilized. These models are expressed in the HSV color model for more easily removable shadows and adaptive illumination changes. Using these models for background subtraction the object areas from a background image can be divided. In addition, some edge images that are generated from an input image and the value of difference between background models and the input image are used. This method can fix the object area size and rule out shadows as a candidate for abandoned objects. Finally we make a decision regarding the abandoned objects on the basis of displacement and presence time. Preliminary experiments show that the present method is effective for detecting abandoned objects.
152.
Tadahiro Oyama, Hillary Kipsang Choge, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Improvement Algorithm for Approximate Incremental Learning, Proc. of ICONIP, 520-529, Bangkok(Thailand), Dec. 2009.
(Summary)
This paper presents an improved algorithm of Incremental Simple-PCA. The Incremental Simple-PCA is a fast incremental learning algorithm based on Simple-PCA. This algorithm need not hold all training samples because it enables update of an eigenvector according to incremental samples. Moreover, this algorithm has an advantage that it can calculate the eigenvector at high-speed because matrix calculation is not needed. However, it had a problem in convergence performance of the eigenvector. Thus, in this paper, we try the improvement of this algorithm from the aspect of convergence performance. We performed computer simulations using UCI datasets to verify the effectiveness of the proposed algorithm. As a result, its availability was confirmed from the standpoint of recognition accuracy and convergence performance of the eigenvector compared with the Incremental Simple-PCA.
153.
Hillary Kipsang Choge, Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Palm print Recognition Based on Local DCT Feature Extraction, Proc. of ICONIP, 639-648, Bangkok(Thailand), Dec. 2009.
(Summary)
In this paper we present a method which extracts features from palmprint images by applying the Discrete Cosine Transform (DCT) on small blocks of the segmented region of interest consisting of the middle palm area. The region is extracted after careful preprocessing to normalize for position and illumination. This method takes advantage of the well known capability of the DCT to represent natural images using only a few coefficients by performing the DCT on each block. After ranking the coefficients by magnitude and selecting only the most prominent, these are then concatenated into a compact feature vector that represents each palmprint. Recognition and verification experiments using the PolyU Palmprint Database show that this is an effective and efficient approach, with a recognition rate above 99 % and Equal Error Rate (EER) of less than 3 %.
154.
Satoru Tsuge, Daichi Koizumi, Minoru Fukumi and Shingo Kuroiwa : Speaker verification method using bone-conduction and air-conduction speech, Proceedinngs of 2009 International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS 2009), 449-452, Kanazawa, Japan, Dec. 2009.
(Summary)
Recently, some new sensors, such as bone-conductive microphones, throat microphones, and non-audible murmur (NAM) microphones, besides conventional condenser microphones have been developed for collecting speech data. Accordingly, some researchers began to study speaker and speech recognition using speech data collected by these new sensors. We focus on bone-conduction speech data collected by the bone-conductive microphone. In this paper, we first investigate speaker verification performances of bone-conduction speech. In addition, we propose a method of using bone-conduction speech and air-conduction together for the speaker verification. The proposed method integrates the similarity calculated by air-conduction speech model and similarity calculated by bone-conduction speech model. Using 99 female speakers' speech data, we conducted speaker verification experiments. Experimental results show that the speaker verification performance of bone-conduction is lower than that of air-conduction speech. However, the proposed method can improve the speaker verification performance of bone- and air-conduction speech. Actually, the proposed method can reduce the equal error rate of air-conduction speech by 16.0% and the equal error rate of bone-conduction speech by 71.7%.
155.
Tadahiro Oyama, Choge H.K., Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Wrist EMG Signals Identification using Neural Network, Proc. of IECON 2009, 4322-4326, Porto, Nov. 2009.
(Summary)
Recently, researches of artificial arms and pointing devices using electromyogram (EMG) have been actively done. However, EMG is usually measured from a part with comparatively big muscular fibers such as arms and shoulders. Therefore, if we can recognize wrist motions using EMG which was measured from the wrist, the range of application will extend furthermore. Moreover, it is predicted that convenience in putting on and taking off the electrode improves. Therefore, we focus on EMG measured from the wrist. In this paper, we aim the construction of wrist EMG recognition system by using fast statistical method and neural network.
Shin-ichi Ito, Yasue Mitsukura, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : A Study on Relationship between Personal Feature of EEG and Human's Characteristic for BCI Based on Mental State, Proc. of IECON 2009, 4265-4268, Porto, Nov. 2009.
(Summary)
This paper discusses the relationship the result classified the electroencephalogram (EEG) patterns while listening to music and the humans nature, which indicates the personal feature of a human, based on the egogram pattern. The EEG analysis calculates the power spectra of the frequency of the EEG signal, divides into the frequency bands based on theta, alpha, and beta rhythms, and evaluates whether the music matches mood of the user or not through EEG pattern classification. A k-nearest neighbor classifier is used to classify the EEG patterns. The egogram is used for detecting nature of the human. Finally, we discuss the relationship the result of EEG pattern classification and the humans nature. An interesting finding was that the recognition accuracy of the EEG pattern meaning the response of them on negative stimuli became high when the subject was classified into the egogram pattern with introverted nature.
157.
Stephen Githinji Karungaru, Sugizaki Masakazu, Minoru Fukumi, Yasue Mitsukura and Norio Akamatsu : Out-of-Focus Blur Image Restoration using the Akamatsu Transform, Proc. of IECON, 4293-4297, Porto(Portugal), Nov. 2009.
(Summary)
In this paper, a new image restoration approach using the Akamatsu transform is presented. In the method, by using repeated simple calculations based on the Akamatsu transform, an out-of-focus blurred image can be restored and sharpened. Out of focus is a major problem bothering many people in photography, especially the amateurs. There exits some solution to this problem on both the hardware and software sides. However, none offers a perfect solution. The Akamatsu transform can easily be embedded into hardware to offer faster processing due to its simplicity. Although, initially proposed for speech processing, this paper show the effectiveness of the transform in image processing. The Akamatsu transform is a combination of integral and differential transforms. The algorithm can be effective tools for image restoration in realtime image processing.
158.
Kohki Abiko, Fukai Hironobu, Yasue Mitsukura, Minoru Fukumi and Masahiro Tanaka : Face Detection Using RBF Network and particle Filter for AIBO, Proc. of the 41st ISCIE International Symposium on Stochastic Systems Theory and Its Applications, Kobe, Nov. 2009.
159.
Hironori Takimoto, Seiki Yoshimori, Yasue Mitsukura and Minoru Fukumi : An Analysis of Influence of Facial Feature for Apparent Age Estimation, Proc. of IIH-MSP 2009, 728-731, Kyoto, Sep. 2009.
(Summary)
The purpose of this paper is to analyze apparent age feature based on age perception. In order to construct the apparent age database, age estimation experiment is performed to the HOIP facial database by 60 subjects. We defined 46 age features which contribute to the age perception from facial image. This process is performed based on research on age acknowledgment in field psychology and technological conventional method about real-age estimation from facial image. The apparent age feature used potentially when human performs age estimation is decided from 46 defined features by using the AIC and the multiple linear regression analysis. By using the proposed method, facial features that influenced on the apparent age in each gender was confirmed.
Seiki Yoshimori, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : An Important Facial Parts Analysis for Evaluating the Impression - Difference by Gender, Proc. of IIH-MSP 2009, 744-747, Kyoto, Sep. 2009.
(Summary)
The concern for image processing has increased very much, in recent year. However, the evaluation of facial impression is not treated in this area that we are naturally doing in our daily life to see on one's face. Wide ranges of applications are released from the achievement of the impression automatic evaluation. Then we analyze important facial parts for impression analysis, in order to achieve automatic evaluation of facial impression. In order to analyze the important facial parts, we performed computer simulation. As the result of this simulation, difference of important areas by gender that relate impression words were detected.
Fukai Hironobu, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : Age and Gender Estimation System Based on Human Perception, Proc. of ROMAN'09, Toyama, Sep. 2009.
(Summary)
In this paper, we propose age and gender estimation system by vairous features. Age and gender has a lot of characteristics. These characteristics are one of the difficult cognitive process in human interaction. If we can extract the important feature of this cognitive process, it is considered that the age and gender estimation by the machine becomes possible. Therefore, we propose a method of age and gender feature extraction and estimation using the face image. In this paper, the age of the face means apparent-age that is based on the human perception of age. Moreover, person's aging and gender difference appear in the faces. For example, the pigmented spot, the wrinkle, sagging skin, shape, color of skin, and so on. Thus, we extract these several features for age and gender estimation. Furthermore, we estimate a continuous age and gender using a neural network (NN).
Yasue Mitsukura, Hironobu Fukai and Minoru Fukumi : Driver Dozing Detection System Using the Near-Infrared Camera Images, Proc. of ICCAS-SICE, 4998-5001, Fukuoka (Japan), Aug. 2009.
(Summary)
Recently, the research to handle face information with computer is being done. Therefore, how to search the face area automatically is very important. For this problem, there are many problems to get the faces. However these researches are for color image. There are few researches using the near-infrared camera. The purpose of this research is to recognize a face with a near-infrared camera. The face detection that used images from near-infrared camera is comparatively difficult to be done, because they are gray scale images. In this paper, the filter by using GA is designed, and the method of detecting the face and the position from the near-infrared images is proposed. It is demonstrate that our approach is effective for vehicle driver monitoring.
163.
Stephen Githinji Karungaru, Sugizaki Masakazu and Minoru Fukumi : Biped Robot Walking Control using Image Processing, Proc. of ICCAS-SICE, 4020-4024, Fukuoka (Japan), Aug. 2009.
(Summary)
In this paper, a vision system for controlling an autonomous biped robot is presented. Robots need sensors to understand circumstances. There are several types of sensors. Recently, a camera is rapidly becoming the sensor of choice due to the remarkable improvement in computer technology especially in processing power. However, there is the blurred image problem because of camera shake due to robot's walking. Our method focuses on the use of image processing to control a humanoid robot's walking. We fix a camera on a biped robot, and input the images from the camera to PC for processing. The results of the image processing are sent to the robot for control immediately. These real-time image processing and control systems give us an efficient application on future robots. In the initial experiment the objective is to control robot walking upstairs.
164.
Hillary Kipsang Choge, Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : A Circle-Based Region-Of-Interest Segmentation Method for Palmprint Recognition, Proc. of ICCAS-SICE, 4993-4997, Fukuoka (Japan), Aug. 2009.
(Summary)
This paper presents a novel method for optimal region-of-interest (ROI) segmentation for palmprint feature extraction based on the largest inscribed circle. This ensures that the optimal amount of features can be extracted, unlike square-based methods which exclude a substantial area on the outside region of the palmprint image. After position normalization, the middle portion of the palmprint is searched to determine the center from which the largest inscribed circle can be extracted. The circular area is then unwrapped into a fixed-size rectangular strip which is further preprocessed to remove redundancies and then split into seven equal square sub-images. A layered approach is then adopted during the matching stage where each square is successively matched and polled to produce a matching score. Experiments are performed using the `PolyU Palmprint Database' and results show that this is a viable method for palmprint feature extraction, with a recognition rate of above 90% obtained.
165.
Tadahiro Oyama, Hillary Kipsang Choge, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Identification of Wrist EMG Signals using Dry Type Electrodes, Proc. of ICCAS-SICE, 4433-4436, Fukuoka (Japan), Aug. 2009.
(Summary)
Recently, researches of artificial arms and pointing devices using electromyogram(EMG) have been actively done. However, EMG is usually measured from a part with comparatively big muscular fibers such as arms and shoulders. Therefore, if we can recognize wrist motions using EMG which was measured from the wrist, the range of application will extend furthermore. However, when we use the wrist EMG, there are problems that the individual difference is large and its repeatability is low and so on. In this paper, we aim the construction of wrist EMG recognition system that is robust to these problems.
166.
Junko Murakami, Hironobu Fukai, Yasue Mitsukura and Minoru Fukumi : Automatic Keyword Additional System by Using the GA anfd the Fuzzy Analysis, Proc. of NCSP'09, 577-580, Hawaii, Mar. 2009.
167.
Seiki Yoshimori, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : Analysis of Important Facial part for Facial Impression Analysis, Proc. of NCSP'09, 392-394, Hawaii, Mar. 2009.
168.
Hironobu Fukai, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : An Apparent-Age Estimation System Based on Several features, Proc. of NCSP'09, 341-344, Hawaii, Mar. 2009.
169.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru and Minoru Fukumi : Improvement of Evolutionary Eye Sensing with Buccal Region",, Proc. of NCSP'2009, 463-466, Honolulu, Mar. 2009.
170.
Toshihiro Yoshida, Takahiro Ogawa, Tadahiro Oyama, Stephen Githinji Karungaru, Yasue Mitsukura, Satoru Tsuge and Minoru Fukumi : Relation Between Mental Change and EEG When Performing Trivial Tasks, Proc. of NCSP'2009, 431-434, Honolulu, Mar. 2009.
(Summary)
Recently, due to long time usage of computers and Visual Display Terminals (VDTs), mental stress and related illnesses have increased tremendously in today's contemporary society. If a computer based method to diagnose stress becames available, it would go a long way in reducing stress illnesses. Hence, it is thought that quantification of stress is important. In this paper, we study the relationship between stress and electroencephalogram (EEG) signals. We studied the effect of stress (the pysiological parameter) when the brain is doing trivial tasks by measuring the brain waves of a subject before and after doing such a task. We apply principal component analysis (PCA) and linear discriminant analysis (LDA) to characterize the change in each component, extract it and discriminate using a neural network (NN). The trivial task used in this experiment is asking the subject to pick and move beans from one plate to another using chopstick. As a result, it was found that PCA is an effective analysis method in this research. Therefore, it can be said that even trivial task have an effect on the brain that may contribute to stress.
171.
Kashihara Yasunori, Satoru Tsuge, Kipsang H. Choge, Tadahiro Oyama, Minoru Fukumi and Shingo Kuroiwa : Non-Stationary Noise Robust Speech Recognition Method using Repetitive Phrase, Proc. of NCSP'2009, 221-224, Honolulu, Mar. 2009.
(Summary)
In this paper, we propose a novel noise-robust speech recognition technique using a repetitive phrase. The repetitive phrase means that the same speech command word is repeatedly uttered. Since there is few repetitive phrase with many syllables in daily conversation, it is consider that a incorrect acceptance of speech commands can be reduced. Moreover, we expect that the non-stationary noise does not influence to the same parts of each phrases in repetitive phrase. Hence, it is able to improve speech recognition performance because we can create a clean speech command by using the repetitive phrase. In this paper, we propse the speech recognition method which combines two-lattices. We conducted the speech recognition experiment using three kinds of noise: a buzzer, a whistle, and a telephone ring. From the experimental results, the proposed method improved the speech recognition accuracy under all noise conditions compared to speech recognition without repetitive phrase. Especially, the proposed method improved the speech recognition accuracy from 21% to 63% under the phone noise condition.
172.
Junko Murakami, Shin-ichi Ito, Yasue Mitsukura, Jianting Cao and Minoru Fukumi : Classification of the vision correction by the EEG, Proc. of NCSP'09, 435-438, Hawaii, Mar. 2009.
173.
Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : A Basic Method for Classifying Humans Based on an EEG Analysis, Proc. of ICARCV'2008, 1783-1786, Hanoi, Dec. 2008.
(Summary)
This paper introduces a method for extracting and confirming a human feature by analyzing prefrontal cortex electroencephalogram (EEG) activities on listening to music that user feels matching, no matching his mood and otherwise. The proposed method is EEG sensing and analyzing to monitor the human feature and classify the human with the activity of different frequency bands of the power spectrum of the EEG signal, which is simple and basic. Finally, the performance of the proposed method is evaluated using real EEG data. We confirm to be able to classify the human into three or more groups.
174.
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Incremental Learning Method for Biological Signal Identification, Proc. of 13th International Conference on Biomedical Engineering (ICBME2008), Singapore, Dec. 2008.
Nakaura Kazuhiro, Stephen Githinji Karungaru, Takuya Akashi, Yasue Mitsukura and Minoru Fukumi : Fast statistical learning with Kerne-Based Simple-FDA, Proc. of IEEE International Conference on Signal Image Technology and Internet Based Systems, 333-337, Bali (Indonesia), Dec. 2008.
(Summary)
In this paper, new statistical learning algorithms with kernel function are presented. Recently, iterative learning algorithms for obtaining eigenvectors in the principal component analysis (PCA) have been presented in the field of pattern recognition and neural network. However, the Fisher linear discriminant analysis (FLDA) has been used in many fields, especially face image analysis. The drawback of FLDA is a long computational time based on a large-sized covariance matrix and the issue that the within-class covariance matrix is usually singular. In order to overcome this difficulty, we proposed the feature generation method simple-FLDA which is approximately derived from geometrical interpretation of FLDA. This algorithm is similar to simple-PCA and does not need matrix operation. In this paper, new statistical kernel based learning algorithms are presented. They are extended versions of simple-PCA and simple-FLDA to nonlinear space using the kernel function. Their preliminary simulation results are given for a simple face recognition problem.
Satoru Tsuge, Minoru Fukumi and Shingo Kuroiwa : Specific speakers' speech corpus over long and short time periods, Proc. of oriental COCOSDA, 45-48, Kyoto, Nov. 2008.
177.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru and Minoru Fukumi : Interactive Interface Using Evolutionary Eye Sensing, Proceedings of the 11th IASTED International Conference Intelligent Systems and Control (ISC 2008), 181-186, Orlando, Florida, USA, Nov. 2008.
178.
Stephen Githinji Karungaru, Fukuda Keiji, Minoru Fukumi and Norio Akamatsu : Classification of Fingerprint Images into Individual Classes Using Neural Networks, Proceedings of the 34th Annual Conference of the IEEE Industrial Electronics Society, 1857-1862, Orlando, Florida, USA, Nov. 2008.
179.
Tsukasa Endo, Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : The Music Analysis Method Based on Melody Analysis, Proc. of ICCAS'2008, 2559-2562, Seoul, Oct. 2008.
(Summary)
Recently, we can have large amounts of music thanks to the development of the computer technology. However, as the data of the music becomes larger, it is a hassle to classify the music based on the music content manually. We think that it is necessary to categorize the music automatically based on our mood. In this paper, we propose a novel method to analyze the music automatically based on melody analysis. The proposed method considers music genres as the measure of music analysis. We extract the acoustic features to characterize the music. Then, we classify music using multi-class classifiers based on the support vector machine (SVM).We adopt two approaches to the multi-class classification method. Furthermore, we propose a visualization method to specify the musical structure on the music genres from the classification result. Finally, computer simulations are done by using real music data in order to prove the effectiveness of the proposed method.
180.
Hironobu Fukai, Yuuki Nisie, Kohki Abiko, Yasue Mitsukura, Minoru Fukumi and Masahiro Tanaka : An Age Estimation System on the AIBO, Proc. of ICCAS'2008, 2551-2554, Seoul, Oct. 2008.
(Summary)
In this paper, we propose an age estimation system on the AIBO. AIBO is an entertainment robot produced by SONY co., Ltd.. AIBO has many sensors to get information around itself and moves according to its instinct. This autonomous action is considerably natural. However, it is inadequate to communicate with people. If AIBO can estimate the human age from a face image, it becomes more excellent entertainment robot. Then, we propose the age estimation method on the AIBO by using face image. In this paper, the apparent age feature is extracted by the fast Fourier transform (FFT), and it is selected by the GA. Moreover, the age is estimated by the 1-dimensional SOM. In order to show the effectiveness of the proposed method, we show the simulation examples.
181.
Satoru Tsuge, Osanai Takashi, Makinae Hisanori, Kamada Toshiaki, Minoru Fukumi and Shingo Kuroiwa : Combination method of Bone-conduction Speech and Air-conduction Speech for Speaker Recognition, Proceedings of Interspeech 2008, 1929-1932, Brisbane, Australia, Sep. 2008.
182.
Yosuke Fukada, Yasue Mitsukura and Minoru Fukumi : The Extraction of the Personal Coloration Pattern for Color Design System, Proc. of SCIS&ISIS, 598-603, Nagoya (Japan), Sep. 2008.
183.
Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : A Method for Filtering Frequency of EEG using Real-Coded Genetic Algorithm, Proc. of SCIS&ISIS, 2097-2100, Nagoya (Japan), Sep. 2008.
(Summary)
This paper proposes a method for filtering frequencies of an electroencephalogram (EEG) on EEG analysis with a real-coded genetic algorithm (RGA). An EEG of humans has the different personal feature that contains an individual error and a change of his/her physical and mental conditions at the occasion arises. The frequencies of the EEG analyzed are the components that contain significant and immaterial information and have different importance. We express the different importance through the weighted value on frequencies. These weighted values are through to express personal feature of EEG signal. The proposed method calculates the power spectra of the frequency of the EEG signal, is weighted these frequencies, divides the frequency bands based on theta, alpha1-3, and beta rhythms, and evaluates whether the music matches mood of the user by EEG pattern classification. A real-coded genetic algorithm is used to specify the weighted value of frequency of the EEG. A k-nearest neighbor classifier is used to classify the EEG patterns. Finally, the performance of the proposed method is evaluated using real EEG data by classifying EEG patterns that listen to music matching, no matching mood of the subject, and feeling otherwise.
184.
Hironobu Fukai, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : Feature Extraction System for Age Estimation, Proc. of KES'2008, 458-465, Zagreb (Croatia), Sep. 2008.
185.
Yohei Tomita, Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Takenori Suzuki : Time-series Models of the EEG Wearing Overcorrected Glasses, Proc. of KES'2008, 450-457, Zagreb (Croatia), Sep. 2008.
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Incremental Algorithm of Discriminant Analysis, Proc. of SCIS&ISIS, 640-643, Nagoya (Japan), Sep. 2008.
(Summary)
This paper presents a new incremental learning algorithm named Incremental Simple-FLDA. It is the algorithm that improved Simple-FLDA to update an eigenvector by using incremental data. Simpe-FLDA is an approximation algorithm of the linear discriminant analysis where an eigenvector can be calculated by a simple repeated calculation. Using this proposal algorithm, the eigenvector is approximately obtained whenever incremental data is added one by one. We carry out computer simulations on personal authentication that uses face images by incremental learning to verify the effectiveness of this algorithm. As a result, the effectiveness of proposal algorithm is shown.
187.
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Incremental Learning Method of Simple-PCA, Proc. of KES'2008, 403-410, Zagreb (Croatia), Sep. 2008.
(Summary)
n this paper, we propose an incremental learning algorithm named Incremental Simple-PCA. This algorithm is added an incremental learning function to the Simple-PCA that is an approximation algorithm of the principal component analysis where an eigenvector can be calculated by a simple repeated calculation. Using the proposed algorithm, it allows to update faster the eigenvector by using incremental data. To verify the effectiveness of this algorithm, we carry out computer simulations on personal authentication that uses face images and wrist motion discrimination using wrist EMG by incremental learning. As a result, we can confirm the effectiveness from the aspects of accuracy and a computing time by comparing the Incremental PCA that gave the incremental learning function to the conventional PCA.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka and Minoru Fukumi : Improving the Robustness of Lips Sensing with Evolutionary Video Processing, Proc. of ITC-CSCC2008, 713-716, Simonoseki, Sep. 2008.
(Summary)
In this paper, an effective method is proposed for robust lips sensing. Our objectives are high-speed lips tracking and data acquisition of a talking person in natural scenes. Our approach is based on the Evolutionary Video Processing. This method has a trade-off between accuracy and a processing time. To solve this problem, we proposed automatic Search Domain Control method and implement this method in the Evolutionary Video Processing. The tracking accuracy is improved from 66.3% to 84.9%. The proposed method can recover from occlusion and tracking loss. Comparative experiments are presented to demonstrate the effectiveness and robustness of the proposed method.
189.
Stephen Githinji Karungaru, Minoru Fukumi, Akashi Takuya and Norio Akamatsu : Facial Gesture Simulation for a Single Image using Image Warping, Proceedings of SCIS&ISIS, 631-635, Nagoya, Japan, Sep. 2008.
190.
TADA Ryosuke, Tadahiro Oyama, Satoru Tsuge, Stephen Githinji Karungaru and Minoru Fukumi : Identification of Music According to Singer, Proc. of ICEE'2008, Vol.P-014, 1-5, Okinawa, Jul. 2008.
(Summary)
We propose a singer identification method using a singer model for retrieving the music data. For the speaker models, we use Gaussian Mixture Models (GMMs) and Vector Quantization (VQ) centroids. To evaluate the proposed method, we conducted the speaker identification experiments. Experimental results show that the correct identification accuracy is 65.0% under the condition that the GMMs are used for the speaker models and the 81.9% under the condition that the VQ centorids are used for the speaker models. We also experimented changing the training data to estimate the best parameter for GMM. The correct identification accuracy is 50.0%, under the condition that training data is 1 and the 95.0%, under the condition that training data is 8. Owing to result of GMM, it has been understood that accurate parameter estimation becomes possible by increasing the number of training data.
191.
YOSHIDA Toshihiro, OGAWA Takahiro, Stephen Githinji Karungaru, Satoru Tsuge, Minoru Fukumi and Yasue Mitsukura : Relation between mental change and EEG when doing Trivial Tasks, Proc. of ICEE'08, Vol.O-160, 1-5, Okinawa, Jul. 2008.
(Summary)
Recently, due to long time usago of computers and Visual Display Terminals (VDTs), mental stress and related illnesses has increased tremendously in today's contemporary society. If a computer based method to diagnose stress were available it would go a long way in reducing stress illnesses. Hence, it is thought that quantification of stress is important. In this research, we study the relationship between stress and electroencephalogram (EEG) signals. We studied the effect of stress (the pysiological parameter) when the brain is doing trivial tasks by measuring the brain waves of a subject before and after doing such a task. We then apply simple principal component analysis (Simple-PCA) to characterize the change in each component, extract it and discriminate using a Neural Network (NN). The trivial task used in this experiment is asking the subject to pick and move beans from one plate to another using chopstick. A recognition accuracy of about 80 % was achieved.
192.
Miyoko Nakano, Stephen Githinji Karungaru, Satoru Tsuge, Takuya Akashi, Yasue Mitsukura and Minoru Fukumi : Face Information Processing by Fast Statistical Learning Algorithm, Proc. of WCCI'08, 3228-3231, Hong Kong, Jun. 2008.
(Summary)
In this paper , we propose a new statistical learning algorithm. This study quantitatively verifies the effectiveness of its feature extraction performance for face information processing. Simple-FLDA is an algorithm based on a geometrical analysis of the Fisher linear discriminant analysis. As a high-speed feature extraction method, the present algorithm in this paper is the improved version of Simple-FLDA. First of all, the approximated principal component analysis (learning by Simple-PCA) that uses the mean vector of each class is calculated. Next, in order to adjust within-class variance in each class to 0, the vectors in the class are removed. By this processing, it becomes high-speed feature extraction method than Simple-FLDA. The effectiveness is verified by computer simulations using face images.
Seiki Yoshimori, Yasue Mitsukura and Minoru Fukumi : Importance analysis of face part in face impression, Proc. of NCSP'08, 244-246, Gold Coast, Mar. 2008.
194.
Miyoko Nakano, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Feature Generation for Face Recognition by Fast Statistical Learning Algorithm, Proc. of NCSP08, 84-87, Gold Coast, Mar. 2008.
(Summary)
In this paper,we proposed a new statistical learning algorithm. This paper quantitatively verifies the effectiveness of the feature extraction performance for face information processing. Simple-FLDA is an algorithm based on a geometrical analysis of the Fisher Linear DIscriminant Analysis. As a high-speed feature extraction method, this algorithm is improved version of Simple-FLDA. First of all, the approximated proncipal component analysis (learing by Simple-PCA) that uses the mean vector of each class is calculated. Next, to adjuast within-class variance in the class to 0, the vectors in the class are removed. By this processing, it becomes hi-speed feature extraction method than Simple-FLDA. The effectiveness is verified by computer simulations using face images.
195.
Hironori Takimoto, Tsubasa Kuwano, Hinorobu Fukai, Yasue Mitsukura and Minoru Fukumi : An Analysis of the Influence of Facial Feature on Age Perception, Proc. of NCSP08, 25-28, Gold Coast, Mar. 2008.
196.
Kipsang H. Choge, Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : Enhancement of Palmprint Images using an Optimized Hexagonal Mulitilayer Perceptron Neural Network, Proc. of NCSP'2008, 423-426, Gold Coast, Mar. 2008.
(Summary)
The image capture and storage stages of palm and other biometric image applications is susceptible to various forms of noise due to varying lighting conditions and the type of image-input process employed. In this paper, we present a method for improving palmprint image contrast and detail using hexagon-shaped fully connected feedforward neural network with a single hidden layer, similar to the ones that have been used in literature for prediction and regression applications. It is trained, using Outputweight- optimized Back Propagation (OWO-BP), to remove various forms of noise added to the input images, producing an image where the fine details of the principal lines and wrinkles are highlighted. The training is performed using sub-images of the 32x32 pixel gray-scale images of the central palm area. This method offers a viable, fast-learning and one-stop alternative to the application of spatial and spectral image processing techniques such as unsharp masking, spatial filtering and FFT-based sharpening algorithms. The network can also be applied to image restoration problems, if the noisy input image is viewed as a degraded version of the target, enhanced image.
197.
Obayashi Katsuyuki, Satoru Tsuge, Minoru Fukumi, Seiji Tsuchiya, Ryosuke Sumitomo, Fuji Ren and Shingo Kuroiwa : A Study of speaker identification using phoneme-information, NCSP'08, 164-167, Gold Coast, Mar. 2008.
198.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka and Minoru Fukumi : Edge Detection of Eye Region using Genetic Algorithm, Proc. of NCSP'08, 88-91, Gold Coast, Mar. 2008.
(Summary)
In this paper, we present a new edge detection method of eye region from a color video sequence. Our purpose is to develop the practical interface using eye movement. In the proposed method, at first, a eye is tracked by the genetic eye tracking method, then, a edge of the palpebral fissure is detected. Difficulty for this detection is determination of a threshold of segmentation of the palpebral fissure and its surrounding region. We try to determine dynamically this threshold by the proposed method. The experimental results shows that the eye is tracked by the genetic eye tracking, and the proposed method can dynamically perform thresholding to detect the palpebral fissure edge. The processing time is almost real-time.
199.
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Fast incremental learning algorithm in pattern recognition, Proc. of NCSP'2008, 196-199, Gold Coast, Mar. 2008.
(Summary)
In a field of pattern recognition, researches on feature extraction and dimension reduction are actively done by using methods of the principal component analysis (PCA) and the linear discriminant analysis (LDA), etc. to data with a lot of attributes. On the other hand, there is an algorithm named Simple- PCA to solve such problems. This is an approximation algorithm of PCA. The algorithm that added the learning function to this Simple-PCA has not been presented yet. In this paper, we propose an Incremental Simple-PCA that added an incremental learning function to the Simple-PCA. Moreover, we carry out the computer simulation on personal authentications that use face images by incremental learning to verify the effectiveness of the proposal algorithm.
200.
Yasuyuki Nakamura, Hironobu Fukai, Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : The Proposal of the GA with Sex-determination of the Hymenoptera and Its Applications, Proc. of NCSP'08, 443-446, Gold Coast, Mar. 2008.
201.
Tsukasa Endo, Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : The Music Classification Method by Melody Analysis Using the SVM, Proc. of NCSP'08, 240-243, Gold Coast, Mar. 2008.
(Summary)
Recently, users can have large amounts of music thanks to the development of the computer technology. However, it becomes hard to identify the music based on the music content manually. We think that it is necessary to categorize the music automatically. We propose the music classification method by melody analysis using the SVM in this paper. We aim at the accomplishment of the automatic music classification. Finally, computer simulations are done by classifying real music data in order to show the effectiveness of the proposed method.
202.
Hironobu Fukai, Hironori Takimoto, Yasue Mitsukura, Toshihisa Tanaka and Minoru Fukumi : Apparent Age Feature Extraction by Empirical Mode Decomposition, Proc. of NCSP'08, 21-24, Gold Coast, Mar. 2008.
203.
Yohei Tomita, Shin-ichi Ito, Yasue Mitsukura, Naoko Koda, Jianting Cao and Minoru Fukumi : Benefits of the Animal Assisted Therapy -- Comparison the EEG features, Proc. of NCSP'08, 299-302, Gold Coast, Mar. 2008.
(Summary)
The animal assisted therapy (AAT) has been known to have psychological and social effects on human being. However, scientific research is insufficient yet. Because, it is difficult to have done in the medical clinic. Furthermore, government can not allow as the medical care because of nothing of clear proof. Therefore, in order to show a scientific basis of the AAT effects, we analyze the electroencephalogram (EEG). Previous study using the stuffed dog has a problem that healing effects from the stuffed dog are little. Therefore, we used a real friendly dog at this experiment to give more healing effects on the subjects. As the results, mental conditions of subjects changed and the brain-wave patterns is confirmed. Finally in this paper, we show the relation between the EEG and mental stability.
204.
Stephen Githinji Karungaru, Minoru Fukumi, Takuya Akashi and Norio Akamatsu : Multiple Faces Detection in Real Time using Neural Networks, Proceedings of the 6th International conference on Computational Intelligent, Man Machine Systems and Cybernetics, 63-68, Canary Islands, Spain, Dec. 2007.
205.
Junko Murakami, Shin-ichi Ito, Yasue Mitsukura, Jianting Cao and Minoru Fukumi : The Proposal of the EEG Characteristic Extraction Method by Using the FCM, Proc. of MJISAT'2007, T4-4-1-T4-4-6, Kuala Lumpur, Nov. 2007.
(Summary)
In this paper, we classify the human conditions (before meal, after meal, before smoking, and after smoking). Moreover, we extract the frequency characteristic of conditions. First of all, we measure the EEG data for subjects by using the simple electroencephalograph. Then, the EEG feature is extracted by using the singular value decomposition (SVD). We classify the human conditions by using the fuzzy cmeans (FCM). Moreover, we perform questionnaires for subjects, in order to analyze the data. Then, in order to show the effectiveness of the proposed method, computer simulations are done.
206.
Koji Sakamoto, Hironobu Fukai, Seiki Yosimori, Yasue Mitsukura and Minoru Fukumi : Image Classification Method Using Evolutionary Regional Segmentation, Proc. of MJISAT'2007, T4-3-1-T4-3-4, Kuala Lumpur, Nov. 2007.
(Summary)
In this paper, we propose the automatic image segmentation method for scene analysis based on color features. The search engines for keyword image retrieval are used on website. However, these search engines usually have some problems. For example, their retrieve images are not suitable image that we don't want to get. Therefore, it is necessary to recognize the images for improving the accuracy of searching. The image segmentation method is considered as helping to recognize the images. The problem of the image segmentation is that the user must decide the thresholds for segmentation. Therefore, the genetic algorithm is used to decide the thresholds for the image segmentation. A supervised image is made based on one of the scene images. Using the color feature values from supervised image, more than 100 test images are segmented into some regions well.
207.
Hayato Kimura, Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : The Sound Emphasis Using the Double Structured ICA, Proc. of MJISAT'2007, T3-4-1-T3-4-4, Kuala Lumpur, Nov. 2007.
(Summary)
Recently, an independent component analysis (ICA) is paid to attention as a method for classifying an independent signal from assembly of different signals. There are a lot of studies using the ICA, for example, the field of image processing, speech recognition, and electroencephalogram (EEG) analysis. Meanwhile, a present musical notation is made by hearing the sound with a person's ear. However, the musical notation is so difficult, because each subject result of the musical notation is different, and the musical notation has many costs. Therefore, we propose a sound emphasis method using the double structured ICA. Then, we aim to obtain the emphasized sounds, for example, the high pitch sound and the low pitch sound. Furthermore, in order to show the effectiveness of the proposed method, we show the computer simulations by using the real data.
208.
Taro Shibanoki, Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Biological Signal Recognition by Fast Statistical Learning Algorithm, Proc. of MJISAT'2007, T2-4-1-T2-4-6, Kuala Lumpur, Nov. 2007.
(Summary)
This paper presents an improved statistical learning algorithm, which is approximately derived from geometrical interpretation of the Fisher linear discriminant analysis (FLDA). Generally, FLDA has to carry out minimization of a within-class variance. However the inverse matrix of the within-class covariance matrix could not be obtained. In order to overcome this difficulty, a new iterative feature generation method, the simple-FLDA was proposed by authors. In this paper, a further improvement algorithm is introduced into the simple-FLDA and its effectiveness is demonstrated for biological signal recognition problems.
209.
Eriko Yokomatu, Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Jianting Cao : An Individual Preference Detection System Using the EEG, Proc. of MJISAT'2007, T2-4-1-T2-4-4, Kuala Lumpur, Nov. 2007.
(Summary)
There are a lot of reports concerning the relation between the EEG and the preference in the analysis of the biosignal. Moreover, attention is currently focused on the aroma, for example, aromatherapy. In this paper, personal preference using olfactory stimulus and the EEG are taken out. We propose the extraction method of the EEG features by the factor analysis. We measure the EEG when smelling favorite odor and dislikeable odor by the simple electroencephalograph. Then, the feature data of the EEG are extracted by the proposed method. Furthermore, in order to show the effectiveness of the proposed method, we demonstrate simulation examples. From the simulation result, it was confirmed that the proposed method works well.
210.
Yosuke Fukada, Keiko Sato, Yasue Mitsukura and Minoru Fukumi : The Extraction of the Coloration Pattern for Personal Room Design, Proc. of MJISAT'2007, T2-3-1-T2-3-5, Kuala Lumpur, Nov. 2007.
(Summary)
We proposed a new type decision support system of modeling the room designs suited for personal preference, and extract the coloration pattern from personal room designs. The color design is one of the most important elements that influence the deciding the image of products. We always pay attention to the coloration patterns, the preference to colors is differed to individual personality. Nevertheless, modeling the room design is depended on the designer's sensitivity. Therefore, the technique of modeling the various room designs people satisfy is needed. In this paper, we propose automatic modeling system of the room design using interactive genetic algorithm (IGA). The proposed system can help a decision support of modeling the various room designs, and reduce the burdens of the design evolution. We simulated the proposed system against some subjects. In addition, from the designs data of the simulation, we extract the each subject's coloration pattern.
211.
Ai Ikeda, Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Character Recognition of Malaysian Vehicle License Plate by Neural Networks, Proc. of MJISAT'2007, T2-3-1-T2-3-4, Kuala Lumpur, Nov. 2007.
(Summary)
This paper presents a method to recognize characters of vehicle license plate in Malaysia by using image processing techniques and a neural network (NN). Vehicle license plate recognition is one of important techniques that can be used for the identification of vehicles all over the world. There are many applications such as entrance admission, security, parking control, airport or harbor cargo control, road traffic control, speed control, toll gate automation and so on. In this paper, techniques for a dynamic image analysis including vehicle detection, license plate segmentation, and character extraction are utilized, which include peculiar knowledge on Malaysian license plate. NN is then used for separated character recognition. The training of NN is carried out using features with boundary coding and gradient components. Computer simulations show that character segmentation and recognition can be effectively carried out using the present method.
212.
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Construction of Online Tuning System based on Wrist EMG, Proc. of MJISAT'2007, T1-7-1-T1-7-5, Kuala Lumpur, Nov. 2007.
(Summary)
Recently, studies of artificial arms and pointing devices using ElectroMyoGram(EMG) have been actively done. However, the individual variation of EMG is large, and its repeatability is low. Furthermore, EMG is usually measured from a part with comparatively big muscular fibers such as arms and shoulders. Therefore, if we can recognize wrist operations using EMG which was measured from the wrist, the range of application will extend furthermore. In this research, we aim toward the development of a device of wristwatch type that consolidates operational interface of various equipments. In the previous research, we proposed a wrist motion recognition system by using Simple-PCA and Simple-FLDA to reduce the number of inputs to a neural network. However, the recognition accuracy was changed considerably by the individual. Therefore, it is thought that the construction of an online tuning system that can consider the individual variation is necessary. In this paper, we propose an improvement algorithm of Simple-PCA named Incremental Simple-PCA that updates eigenvectors sequentially by adding a new data. Moreover, we try the construction of the online tuning system by using this algorithm.
213.
Shin-ichi Ito, Hiroko Nakamura, Yasue Mitsukura, Takafumi Saito and Minoru Fukumi : A Visualization of Genetic Algorithm Using the Pseudo-color, Proc. of ICONIP'2007, Vol.2, 444-452, Kitakyushu, Nov. 2007.
(Summary)
In this paper, we propose the method for determining the schema interactively based on a visualization method to grasp the search process and results in the binary-coded genetic algorithm. We focus on the chromosome structure, the fitness function, the objective function, and the association among these parameters, because these parameters are decided interactively and very difficult to disentangle their effects. We can indicate the most important parameters optimized in visually. The visualization method is indicated all individuals of the current generation using the pseudo-color. Moreover, the visualization method proposed uses to support for determining the schema. In order to show the effectiveness of the proposed method, we apply the proposed method to the zero-one knapsack problems.
214.
Yosuke Fukada, Keiko Sato, Yasue Mitsukura and Minoru Fukumi : The Room Design System of Individual Preference with IGA, Proc. of ICCAS'2007, 2158-2161, Seoul, Oct. 2007.
(Summary)
In this paper, we illustrate a decision support system of modeling the room designs suited for personal preference. The preference to colors is differed to individual personality. Nevertheless, modeling the room design is depended on the designer's sensitivity until now. Therefore, the technique of modeling the various room designs people satisfy is needed. In this paper, we propose automatic modeling system of the room design using interactive genetic algorithm (IGA). IGA has the advantage of learning the user's interactive evaluation, which is applied to the production of new room designs suited for the personal preference. The proposed system can help a decision support of modeling the various room designs, and reduce the burdens of designers significantly. In proposed system, the design evolution of IGA can reflect the user preference, and we present the effectiveness of the system by simulations.
215.
Hironobu Fukai, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : An Apparent Age Estimation System Using the Evolutionary Algorithm, Proc. of ICCAS'2007, 2146-2149, Seoul, Oct. 2007.
(Summary)
The age is one of important information in our living. If the age estimation that uses face image by computer becomes possible, it is thought that the age estimation assumes an important role in various scenes. In this paper, we propose an age estimation by using the supervised SOM. Further, the important features for the age estimation are selected by the GA.
216.
Shin-ichi Ito, Yasue Mitsukura, Jianting Cao and Minoru Fukumi : Detecting Method of Music to Match the User's Mood in Prefrontal Cortex EEG Activity Using the GA, Proc. of ICCAS'2007, 2142-2145, Seoul, Oct. 2007.
(Summary)
In this paper, we propose a method for detecting the mood much music for prefrontal cortex electroencephalogram (EEG) activity. The analyzed EEG frequencies contain significant and immaterial information components. We focused on the combinations of the significant frequency. These frequency combinations are thought to express personal features of EEG activity. In the proposed method, we calculate the spectrum of these frequency combinations rates that does not include the noise frequency components and evaluates whether the music matches the user's mood through a simple threshold processing. Then, a genetic algorithm (GA) is used to specify the frequency of personal features on the EEG. The threshold vale used the threshold processing is the average value of the spectrum rates specified EEG frequency combinations. Finally, the performance of the proposed method is evaluated using real EEG data.
217.
Junko Murakami, Shin-ichi Ito, Yasue Mitsukura, Jianting Cao and Minoru Fukumi : Detection of the Human-Activity Using the FCM, Proc. of ICCAS'2007, 1995-1998, Seoul, Oct. 2007.
(Summary)
In this paper, we propose the detection system of the human activity by using the electroencephalograms (EEG). First, we measure the EEG data for subjects. In most of all conventional studies, the EEG having a lot of sensors is used. Therefore, subjects must eat or smoke while using the EEG interface. However, this situation is not practical for subjects. In this study, taking account of the burden of subjects, we use only one measurement point 'FPI'. First, we measure the EEG data and the EMG data for subjects. Then, the EEG feature is extracted by using the singular value decomposition (SVD). From the result, we classify the EEG pattern by the fuzzy c-means (FCM). If we cannot classify the EEG pattern into each activity, the discriminant analysis (DA) is used. We consider the EEG features of activities. Then, in order to show the effectiveness of the proposed method, computer simulations are done.
218.
Koji Sakamoto, Hironobu Fukai, Yasue Mitsukura and Minoru Fukumi : Automatic Decision Method of Parameters in the Maximum Distance Algorithm, Proc. of ICCAS'2007, 785-788, Seoul, Oct. 2007.
(Summary)
The maximum distance algorithm has been considered to be effective for an image segmentation of color scenery images. However, in the maximum distance algorithm, the parameter which decides the end term of the clustering is set in advance. The applicable value of this parameter depends on the individual image. Therefore, we propose the automated adjustment method of the maximum distance algorithm's parameter for the image segmentation of scenery images. First, "image density" is defined as the measure to evaluate the complexity of each image. Image density is calculated by difference between average of color value and color value of each pixel. Then, the relation of the image density and applicable value of the maximum distance algorithm is investigated. This investigation enables us the automated adjustment method of maximum distance algorithm's parameter fitting the image density of individual image. In this paper, the computer simulation is done for the purpose of comparing the conventional method and proposed method. There is the regulation between appropriate parameter in maximum distance algorithm. The experiment with about 100 images shows the effectiveness of the proposed method.
219.
Shin-ichi Ito, Yasue Mitsukura, Jianting Cao and Minoru Fukumi : Method for Detecting Music to Match the User's Mood in Prefrontal Cortex Electroencephalogram Activity Based on Individual Characteristics, Proc. of 2007 IEEE International Conference System, Man and Cybernetics, 2640-2644, Montreal, Oct. 2007.
(Summary)
In this article, we propose a method for detecting music to match a users mood in prefrontal cortex electroencephalogram (EEG) activity. The frequencies of the EEG analyzed are the components that contain significant and immaterial information. We focused on the significant frequency combinations. These frequency combinations are thought to express individual characteristics of EEG activity. The proposed method calculates the percentage of the spectrum of these frequency combinations that does not include the noise frequency components and evaluates whether the music matches the users mood through a simple threshold processing. A genetic algorithm (GA) is used to specify the frequency of individual characteristics on the EEG. A threshold value that used the threshold processing is determined in the GA. Finally, the performance of the proposed method is evaluated using real EEG data.}
Shin-ichi Ito, Yasue Mitsukura, Hiroko Miyamura, Takafumi Saito and Minoru Fukumi : The EEG Feature Extraction Method of Listening to Music Using the Genetic Algorithms and Latency Structure Model, Proc. of SICE'2007, 2823-2826, Takamatsu, Sep. 2007.
(Summary)
It is known that an electroencephalogram (EEG) is characterized by the unique and personal features of an individual. The EEG frequency components are contained the significant and immaterial information, and then each importance of these frequency components is different. These combinations are often unique like individual human beings and yet they have underlying basic characteristics. We think that these combinations and/or the importance of the frequency components show the personal features. Therefore we propose the two techniques for estimating the personal features. A simple genetic algorithm is used for specifying these frequency combinations. Other technique, a real-coded genetic algorithm is used for estimating the importance of EEG frequency components. Then a latency structure model based on the personal features is used for extracted the feature vector of the EEG. Furthermore, the visualization map is used for evaluating the extracted feature vector of the EEG. In order to show the effectiveness of the proposed methods, the performance of the proposed method is evaluated using real EEG data.
221.
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Incremental learning method in wrist EMG recognition system, Proc. of SICE'2007, 2819-2822, Takamatsu, Sep. 2007.
(Summary)
In a field of pattern recognition, researches of feature extraction and dimension reduction using the Simple-PCA that is an approximation algorithm of the principal component analysis (PCA) are actively conducted. In such a statistical method, a lot of algorithms that perform incremental learning by using new incremental data exist. For example, there is an algorithm named Incremental PCA for PCA. However, the algorithm capable of performing incremental learning to the Simple-PCA has not been presented yet. In this paper, we propose Incremental Simple-PCA that introduces an incremental learning function to the Simple-PCA. Moreover, the effectiveness of this algorithm is verified in EMG pattern recognition.
222.
Hironori Takimoto, Tsubasa Kuwano, Yasue Mitsukura, Hironobu Fukai and Minoru Fukumi : Appearance-age feature extraction from facial image based on age perception, Proc. of SICE'2007, 2813-2818, Takamatsu, Sep. 2007.
(Summary)
An age is relatively important information in the various features which are shown human from face. It is possible to apply it to an amusement and an aesthetic surgery by analyzing the appearance-age feature which is used for age perception in a facial region. In this paper, we analyze facial feature of appearance-age based on age perception. In order to construct the appearance-age database, age estimation experiment is performed to the HOIP facial database by 15 subjects. The appearance-age feature used potentially when human performs age estimation is decided by using the genetic algorithms and the neural network. By using the proposed method, facial features that influenced on the appearance-age in each gender was confirmed. Moreover, it is suggested that that appearance-age feature which is used for age estimation is different in each generation when human performs age estimation.
223.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Proposal of Adaptive Graininess Suppression Method, Proc. of SICE'2007, 2827-2831, Takamatsu, Sep. 2007.
(Summary)
Previous studies of image restoration for noise image were based on mask processing. These conventional noise removal methods represented from mask processing have issue of definition degradation to accompany spacial processing. In this paper, we propose a graininess suppression method based on edge shape. In this method, we detect edges from a noise image and perform graininess suppression for this image based on edge information. On the edge detection, we execute an image transformation for an image that enables us to extract edge by making principal component image. Moreover, we use the canny edge detection operator that can detect a weak edge that relates to a real edge, and do not detect a lie edge. In the suppression process, we use Wiener filter that can restore an noise image without making a complete edge map and the original signal map. We have that the present method for the noise added images to verify effectiveness and have confirmed this.
224.
Junko Murakami, Shin-ichi Ito, Yasue Mitsukura, Jianting Cao and Minoru Fukumi : A Design of the EEG feature detection and condition classification, Proc. of SICE'2007, 2798-2803, Takamatsu, Sep. 2007.
(Summary)
In this paper, we classify the human conditions (before and after meal, before and after smoking) and extract the frequency feature of conditions by using the electroencephalograms (EEG). First, we measure the EEG data. Then, we classify the conditions by using the principal component analysis (PCA). Moreover, the EEG data is reconstructed by using the questionnaires and the result of classification. From the result, we consider ideal circumstance for the EEG measurement. Finally, the EEG data is decompressed to consider the EEG features of conditions. Then, in order to show the effectiveness of the proposed method, computer simulations are done.
225.
Hironobu Fukai, Hironori Takimoto, Yasue Mitsukura and Minoru Fukumi : Apparent age estimation system based on age perception, Proc. of SICE'2007, 2808-2812, Takamatsu, Sep. 2007.
(Summary)
In this paper, we propose the new age estimation system with the GA and the LVQ. The GA is used in the feature extraction of faces at each age and the faces are estimated by using the LVQ in this paper. Moreover, features are analyzed in the age by using the selected feature data. Furthermore, in order to show the effectiveness of the proposed method, we show the simulation examples. From the simulation results, we can confirmed that the proposed method works well.
226.
Eriko Yokomatsu, Shin-ichi Ito, Yasue Mitsukura, Jianting Cao and Minoru Fukumi : A Design of the Preference Acquisition Detection System, Proc. of SICE'2007, 2804-2807, Takamatsu, Sep. 2007.
(Summary)
Recently in the world, there are a lot of reports concerning the relation between the EEG and the preference in the analysis of the biosignal. In this paper, personal preference using olfactory stimulus and the EEG are taken out. We measure the EEG when smelling favorite odor and dislikeable odor. Then, the feature data of the EEG are extracted by the proposed method. Furthermore, in order to show the effectiveness, we demonstrate simulation examples. From the simulation result, it was confirmed that the proposed method works well.
227.
Minoru Fukumi, Stephen Githinji Karungaru, Satoru Tsuge, Miyoko Nakano, Takuya Akashi and Yasue Mitsukura : Fast Statistical Learning Algorithm for Feature Generatio, Proc. of KES2007, Vol.LNAI 4694, 91-97, Vietri sul Mare, Italy, Sep. 2007.
(Summary)
This paper presents an improved statistical learning algorithm for feature generation in pattern recognition and signal processing. It is approximately derived from geometrical interpretation of the Fisher linear discriminant analysis (FLDA). Recently FLDA has been often used in many fields, especially face image recognition. The drawback of FLDA is a long computational time based on a large-sized covariance matrix and the issue that the within-class covariance matrix is usually singular. In order to overcome the difficulty of FLDA, a new iterative feature generation method, the simple-FLDA was proposed by authors. In this paper, further improvement is introduced into the simple-FLDA and its effectiveness is demonstrated for preliminary personal identification problem.
Mitsuhiro Ozawa, Satoru Tsuge, Masami Shishibori, Kenji Kita, Minoru Fukumi, Fuji Ren and Shingo Kuroiwa : Automatic Utterance Segmentation Tool for Speech Corpus, IEEE NLP-KE2007, 401-406, Beijing, Aug. 2007.
(Summary)
We collect the speech data for investigating an intra-speakers' speech variability over a short and long time. In general, to reduce the load of speakers, the speech data are collected as one file from collecting start to collecting end. Hence, there are some noises, non-speech sections and mistaken sections in this file. Consequently, we must segment this file into individual utterances and select the useful utterances. This process requires a lot of time and efforts. In this paper, we propose an automatic utterance segmentation tool for dividing the collected speech data. The proposed tool is composed of four processes, which are a voice activity detection, speech recognition, a DP matching, and a correct of speech section. For evaluating the proposed tool, we conduct the evaluation experiments using a female speaker's speech data in our corpus. Experimental results show that the proposed method can reduce a filing time by 90% compared to a manual filing. In This paper, first, we introduced the large speech corpus. This speech corpus contains is the speech data collected by specific speaker over long and short time periods. And, we explained the automatic utterance segmentation tool which we made in the case of corpus build. And inspected the validity. As a result, it was demonstrated that the automatic utterance segmentation tool was high-performance. Furthermore, it was demonstrated that speech corpus build became simple by using the automatic utterance segmentation tool.
Satoru Tsuge, Keiji SEIDA, Masami Shishibori, Kenji Kita, Fuji Ren, Minoru Fukumi and Shingo Kuroiwa : Analysis of Variation on Intra-Speakers Speech Recognition Performances, IEEE NLP-KE2007, 387-392, Beijing, Aug. 2007.
(Summary)
Even if a speaker uses a speaker-dependent speech recognition system, speech recognition performance varies. However, the relationships between intra-speaker's speech variability and speech recognition performance are not clear. To investigate these relationships, we have been collecting speech data since November 2002. In this paper, we analyze the relationships between intra-speaker's speech variability and the phoneme accuracy by a correlation analysis. Analyzed results showed the strong negative correlation between the phoneme accuracy and the speaking rate. The correlation coefficient indicated -0.77. Moreover, we can see that the phoneme accuracy is correlated with the temperature in the recording room and the humidity difference.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka and Minoru Fukumi : Genetic Eye Tracking for Blinking, Proc. of the 2007 International Conference on Image Processing, Computer Vision, and Pattern Recognition (IPCV'07), Las Vegas, Jun. 2007.
231.
Takuya Akashi, Toshiya Nishimura, Yuji Wakasa, Kanya Tanaka and Minoru Fukumi : Practical Genetic Eye Detection System: Which Is Better, Optical or Digital Zoom?, Proc. of 2007 RISP International Workshop on Nonlinear Circuits and Signal Processing, 535-538, Shanghai, Mar. 2007.
(Summary)
In this paper, the effectiveness of the digital zooming for the size and orientation invariant eye tracking system in an active scene using template matching and genetic algorithms is described. Consideration the practical use, it is better that the equipment is simple and low cost. In our proposed system, we use a ``web camera'', which requires no specific interface. This type of camera is easily obtainable, however does not have optical zooming function. Therefore, we try to use the digital zooming instead of the optical zooming. However, the quality of image is a problem. In this paper, the effectiveness of the digital zooming for the eye detection is demonstrated. Our proposed system is based on template matching using genetic algorithms (GAs). Only one artificial template is used for every human user. Various geometric changes of iris can be supported by genetic algorithms. As a result of comparative experiments, the eye detection accuracy of the digital zooming is little better than the optical digital zooming, though the target image is coarse by the digital zooming
232.
Hayato Kimura, Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : Music Part Classification Using the ICA, Proc. of 2007 RISP International Workshop on Nonlinear Circuits and Signal Processing, 485-488, Shanghai, Mar. 2007.
(Summary)
Recently, independent component analysis (ICA) is paid to attention as a method for classifying an independent signal from assembly of different signals, and there are a lot of researches using the ICA. Then, in this paper, we pay attention to the field of voice recognition, and we aim at the achievement of the part classification system of music using the ICA. Furthermore, in order to show the effectiveness of the proposed method, we show the simulation examples.
233.
Hironobu Fukai, Yasue Mitsukura, Hironori Takimoto and Minoru Fukumi : An Age Estimation System Based on the LVQ, Proc. of 2007 RISP International Workshop on Nonlinear Circuits and Signal Processing, 449-452, Shanghai, Mar. 2007.
(Summary)
In this paper, we propose the new age estimation system with the GA and the LVQ. The GA is used in the feature extraction of faces at each age and the faces are estimated by using the LVQ in this paper. Moreover, features are analyzed in the age by using the selected feature data. Furthermore, in order to show the effectiveness of the proposed method, we show the simulation examples. From the simulation results, we can confirmed that the proposed method works well.
234.
Tadahiro Oyama, Yuji Matsumura, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Discrimination of Wrist EMG Signals using a Statistical Method, Proc. of 2007 RISP International Workshop on Nonlinear Circuits and Signal Processing, 393-396, Shanghai, Mar. 2007.
(Summary)
In this study, we aim toward the development of a device of wristwatch type that consolidates operational interface of various equipments. In the previous study, the recognition accuracy was changed considerably by the individual. Therefore, we thought that the construction of an online tuning system that can consider the individual variation was necessary. In this paper, we propose an improvement algorithm of Simple-PCA that changes eigenvectors sequentially by adding data. Moreover, the construction of the online tuning system is carried out by using this algorithm.
235.
Eriko Yokomatsu, Shin-ichi Ito, Yasue Mitsukura, Jianting Cao and Minoru Fukumi : Preference Acquisition by Using Olfactory Stimulus and EEG, Proc. of 2007 RISP International Workshop on Nonlinear Circuits and Signal Processing, 381-384, Shanghai, Mar. 2007.
(Summary)
In this paper, personal preference using olfactory stimulus and the EEG are taken out. We propose the extraction method of the EEG features by the factor analysis. Then, we classify the odor into favorite and dislikeable. Furthermore, in order to show the effectiveness of the proposed method, we demonstrate simulation examples. From the simulation result, it was confirmed that the proposed method works well.
236.
Junko Murakami, Shin-ichi Ito, Yasue Mitsukura, Jianting Cao and Minoru Fukumi : Classification of Conditions by the EEG, Proc. of 2007 RISP International Workshop on Nonlinear Circuits and Signal Processing, 297-300, Shanghai, Mar. 2007.
(Summary)
In this paper, we classify the human conditions (before meal, after meal, before smoking, and after smoking) by using the electroencephalograms (EEG).We propose the electroencephalograms (EEG) characteristic extraction method by using the principal component analysis (PCA). Then, in order to show the effectiveness of the proposed method, computer simulations are done by classifying the EEG pattern in the proposed method.
237.
Yosuke Fukada, Keiko Sato, Yasue Mitsukura and Minoru Fukumi : A Design of the Mapping Method of Color and Kansei, Proc. of 2007 RISP International Workshop on Nonlinear Circuits and Signal Processing, 73-76, Shanghai, Mar. 2007.
(Summary)
We have various sensibilities to various colors, and the colors have a great influence on deciding the image of products. We have always paid attention to the colors, the image to colors depends on individual personality. Therefore, the technique of analyzing image to colors is needed in KANSEI (Japanese term) information processing field. In this paper, we investigate the evaluation between colors and images against subjects, and analyze the relationship to individual. We propose the method of the KANSEI extraction from each subject data by using 2 grade-type sand grass neural network. Finally, we describe the KANSEI extraction information visually. Furthermore, in order to show the effectiveness of the proposed method, we show the simulation examples. From the simulation results, it was confirmed that the proposed method works well.
238.
Satoru Tsuge, Minoru Fukumi, Masami Shishibori, Fuji Ren, Kenji Kita and Shingo Kuroiwa : Study of Relationships Between Intra-Speaker's Speech Variability and Speech Recognition Performance, 2006 International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS 2006), 41-44, Tottori, Japan, Dec. 2006.
(Summary)
Even if a speaker uses a speaker-dependent speech recognition system, speech recognition performance varies. For this reason, speech quality is varied by some factors, including emotion, background noise, and so on, even though the speaker and utterance remain constant. However, the relationships between intra-speaker's speech variability and speech recognition performance are not clear. Hence, we focus on the intra-speaker's speech variability which affects the speech recognition performances. To investigate these relationships, we have been collecting speech data since November 2002. Using a part of the speech corpus, we conducted speech recognition experiments. In this paper, we analyze the relationships between intra-speaker's speech variability and the phoneme accuracy by using the correlation analysis. For factors of the correlation analysis, we use a number of errors, a speaking rate, a likelihood. Analysis results show a strong correlation between the number of the substitution errors and the phoneme accuracy although the correlations of the number of the deletion and the insertion errors are low. Therefore, it is considered that there are overlaps between phonemes since the feature parameters vary at each speaking rate. For improving the phoneme accuracy, it is needed that we study a method which discriminates phonemes. On the other hand, although the correlation between the phoneme accuracy and the speaking rate seems to be low, a strong correlation between the speaking rate and the number of deletion errors and insertion errors are found. Since the number of the insertion errors and the number of the deletion errors were in the counterbalance relation, the correlation between the speaking rate and the phoneme accuracy was low. However, we consider that it is needed to normalize the speaking rate because the speaking rate influences on the number of the deletion and the insertion errors.
239.
Kou Nakamichi, Stephen Githinji Karungaru, Minoru Fukumi, Takuya Akashi, Yasue Mitsukura and Motokatsu Yasutomo : Extraction of the Liver Tumor in CT Images by Real-coded Genetic Algorithm, Proc. of IASTED CI'2006, 366-371, San Francisco, Nov. 2006.
(Summary)
The purpose of this work is the construction of an automatic diagnosis support system for CT images in order to reduce the doctor's load. Toward this end, in this paper, a method to extract liver tumors in CT images using a real-coded genetic algorithm is proposed. Conventionally, a threshold is necessary to extract an object from an image. However, such a method is not effective for CT images because gray scale values are different in each image. Therefore, in this paper, we propose the method for extracting the tumor in the liver from the CT image without the need of a threshold. In this method, a polygon enclosure of the liver tumor is extracted using a GA.
240.
Takenaka Yusuke, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Isolated Syllable Recognition on Mobile Terminal Devices, Proc. of IASTED International Conference on Computational Intelligence, 366-371, San Francisco, Nov. 2006.
(Summary)
The aim of this work is the establishment of isolated syllable recognition for Japanese language on mobile terminal devices and/or the retry after the failure in speech recognition. The subjects in this study are five Japanese vowels. The experiment was carried out on 100 sets of five Japanese vowels from four men. The experiment for recognition was carried out using the Akamatsu transform. The experiment demonstrated the effectiveness of the proposed technique for recognizing the isolated 100 sets. This paper serves as an introduction on the use of the Akamatsu transform in speech recognition and does not, as yet, compare the results with existing methods.
241.
Kentaro Tohi, Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : Pattern Recognition of EMG Signals by the Evolutionary Algorithms, Proc. of SICE-ICCAS'2006, 2574-2577, Busan, Oct. 2006.
242.
Satomi Ota, Shin-ichi Ito, Yasue Mitsukura and Minoru Fukumi : Proposal for the Extraction Method of Personal Comfort and Preference by the EEG Maps, Proc. of SICE-ICCAS'2006, 604-607, Busan, Oct. 2006.
243.
Keiko Sato, Yasue Mitsukura and Minoru Fukumi : Designing of the Color KANSEI Information Map Considering the Individual Characteristic by Using Multivariable Analysis, Proc. of SICE-ICCAS'2006, 3706-3710, Busan, Oct. 2006.
(Summary)
The color design is one of the most important elements that influence the impression of products, therefore the technology which understands and reflects the consumer's sensibility is needed in the color design. Especially, the color coordination system that connects colors with impressions is expected as the system supporting the color design. Therefore, in the field of the KANSEI engineering, researches that model the correlation of colors (the arrangement of the color) and impression words have been done. However, the impression received from the color is various by individual, therefore, it is necessary to model the KANSEI information to each individual. In this paper, we analyze the data of the evaluation questionnaire of the color for the subject, and propose the method of extracting the KANSEI information of the subject. Concretely, by using the multivariable analysis, the correlation with an amount of the sensibility and the color feature is found. As a result, it is thought that the tendency of impressions to colors can be expressed in the sight.
244.
Kentaro Tohi, Yasue Mitsukura and Minoru Fukumi : Pattern Recongnition of EMG signals by the Evolutionary Algorithms, Proc. of SICE-ICCAS'2006, 2574-2577, Busan, Oct. 2006.
(Summary)
In this paper, we propose a method of function derivation for performing recognition of wrist operations by the electromyographic (EMG) signals extracted from 4-channel EMG sensor. In designing a recognition device of operations, the important fewer amount of information is needed for reduction of cost and accuracy improvement in practical systems. Then, date mining is performed by specifying important frequency bands using genetic algorithm (GA) and neural network (NN). The derivation of function for generating a feature vector is performed only using the important frequency bands obtained by GA and NN. In this case, the feature vector which consists of frequency spectrum to be used is mapped to another space. We use the generated function as an input feature to perform recognition experiments of EMG signal by NN. Finally, the effectiveness of this method is demonstrated by means of computer simulations.
245.
Hironobu Fukai, Yasue Mitsukura and Minoru Fukumi : A Design of an Age Estimation System Using the SOM, Proc. of SICE-ICCAS'2006, 2582-2585, Busan, Oct. 2006.
(Summary)
In this paper, we propose the new age estimation system using the GA and SOM. The GA is used in the feature extraction of faces at each age. Especially, we use the interactive GA. The faces are classified by using the SOM. Moreover, features are analyzed in the age by the selected parts of faces. Furthermore, in order to show the effectiveness of the proposed method, we show the simulation examples.
246.
Yasue Mitsukura and Minoru Fukumi : Fast Face Extraction from Near-Infrared Camera Images, Proc. of SICE-ICCAS'2006, 5711-5714, Busan, Oct. 2006.
(Summary)
This paper proposes and evaluates the use of digital image processing for profiling a key groove. The first step of research is taking a cross-sectional image of the key. Exploiting Laplacian edge enhancement to acquire a contour of the key grooves. Transform the image data to surface data presented by coordinate (x,y). This information is used as an input for CNC coding program. The duplication of key blank is performed by CNC. The tolerance examination is conducted by comparing a calculated image proposed to the key-groove with a measured value from vernier. The error is less than 5%
247.
Tadahiro Oyama, Yuji Matsumura, Stephen Githinji Karungaru, Yasue Mitsukura and Minoru Fukumi : Recognition of Wrist Motion Pattern by EMG, Proc. of SICE-ICCAS'2006, 599-603, Busan, Oct. 2006.
(Summary)
Recently, studies of artificial arms and pointing devices using ElectroMyoGram (EMG) have been actively done. However, the individual variation of EMG is large, and its repeatability is low. Furthermore, EMG is usually measured from a part with comparatively big muscular fibers such as arms and shoulders. Therefore, if we can recognize wrist operations using EMG which was measured from the wrist, the range of application can extend furthermore. In this study, we aim toward the development of a device of wristwatch type that consolidates operational interface of various equipments. In particular, as an early stage, we propose a wrist motion recognition system. First, we execute the Fourier transform to the signal for feature extraction. Next, we experiment it by using neural networks after the dimensional reduction by using simple-PCA and simple-FLDA to reduce the number of inputs to NN. It was confirmed that the present approach was one of the techniques which were effective in the wrist recognition experiment.
248.
Shin-ichi Ito, Yasue Mitsukura, Hiroko Miyamura, Takafumi Saito and Minoru Fukumi : The Proposal of the EEG Characteristics Extraction Method in Weighted Principal Frequency Components Using the RGA, Proc. of SICE-ICCAS'2006, 1152-1155, Busan, Oct. 2006.
(Summary)
An EEG has frequency components which can describe most of the significant features. These combinations are often unique like individual human beings and yet they have underlying basic features. These frequency components are contained the important and/or not so important components, and then each importance of these frequency components are different. The real-coded genetic algorithm (: RGA) is used for selecting and being weighted the principal characteristic frequency components. We attempt to construct mental change appearance model (: MCAM) of only one measurement point. In order to show the effectiveness of the proposed method, computer simulations are carried out by using real data.
249.
Stephen Githinji Karungaru, Minoru Fukumi, Akashi Takuya and Norio Akamatsu : A Simple 3D Edge Template for Pose Invariant Face Detection, Proc. of Knowledge-Based & Intelligent Information & Engineering Systems (KES), 692-698, Bournemouth, Oct. 2006.
250.
Morioka Yoshiyuki, Stephen Githinji Karungaru, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Automatic Extraction of Heart Contour using SNAKES, Proc. of SICE-ICCAS 2006, 1477-1480, Busan, Oct. 2006.
(Summary)
Recently, as sophisticated medical instruments have been developed, the internal state of a human body has become well-known. In particular, in the examination of heart, non-surgical and non-intrusion ultrasonic diagnostics are generally utilized. However, the diagnosis depends on the doctor's subjectivity and experience. Furthermore, the doctor's burden increases because the contour of the heart is extracted by manual operations. Therefore, the development of a system that offers objective information in addition to a subjective judgment of the doctor by extracting and analyzing quantitative information from image analysis by computers is necessary. In this paper, the contour of the chamber wall is extracted by using an active contour model SNAKES as a preprocessing of the heart disease detection. The contour has been extracted by using SNAKES of the expansion model from the inner surface of heart
251.
Ogawa Takahiro, Stephen Githinji Karungaru, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Feature Extraction in Listening to the Music Using Statistical Analysis of the EEG, Proc. of SICE-ICCAS 2006, 5120-5123, Busan, Oct. 2006.
(Summary)
In order to solve stress problems, researchers have studied healing, especially the music therapy. It is mentioned that objective evaluation of the music therapy is an important assignment, and some researchers have tried objective measurement based on physiological change. In this paper, the purpose is extraction of features that may be influenced by the music. We pay attention to EEG (electroencephalogram) as an objective and absolute scale. This paper proposes a method that extracts features of the EEG by the CDA (canonical discriminant analysis). From the result of the experiment, it is suggested that the CDA extracts the features influenced by the individual and the music type
252.
Watanabe Takumi, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Design of an IR Communication Link for a Computer-Controlled Humanoid Robot, Proc. of SICE-ICCAS 2006, 3496-3499, Busan, Oct. 2006.
(Summary)
Recently, many humanoid robots can be formed on television programs and public events. In this paper, we pay attention to such robots. Under this background, diverse parts have been developed and many symposiums have been held to share domain knowledge with many engineers. Furthermore, as for the robot technology, we think that a robot which makes humanlike movement and general view has been developed. However, if a multifunction robot were developed, it could work instead of a human and our lifes could be more convenient and less risky. However autonomous action is very difficult to achieve because a robot cannot recognize varying environment all of the time. In this paper, we focus on how to recognize such environments. The environment is recognized using a camera and is learned with a neural network (NN). The autonomous action can be learned by the NN on the robot or transferred from a PC after action learning. In this work, we use the latter method, and then transmit the image that is necessary to control the robot from a PC because the robot cannot acquire the image by itself.
253.
Takahashi Yasuyuki, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Feature Point Extraction in Face Image by Neural Network, Proc. of SICE-ICCAS 2006, 3783-3786, Busan, Oct. 2006.
(Summary)
Conventionally, manual operations that specify positions of feature points such as eyes and nose are needed when morphing is carried out for a face image. In this work, the feature points are therefore extracted by using face area detection and a feature points decision methods to automate positional specification of feature points. As a result, the morphing of a face image can be carried out without manually specifying feature points. Face area detection is achieved by a threshold method using the YIQ color system. Feature points decision method extracts feature points by using a 3 layer perceptron type neural network (back-propagation). The attribute of the feature of eyes is defined to be a value of A in the color system LAB. In the same way, the attribute of feature points of the lip is defined as a value of B in the color system LAB. The extraction experiment of feature points was conducted from 120 face images by using the neural network, and the effectiveness of the present method was verified.
254.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru and Minoru Fukumi : Downsized Genetic Algorithm by Automatic Search Domain Control for Lips Detection, Proc. of the International Conference on Soft Computing and Intelligent Systems and International Symposium on advanced Intelligent Systems (SCIS & ISIS 2006), 1312-1317, Tokyo, Sep. 2006.
(Summary)
In this paper, a method of downsizing a genetic algorithm (GA) for lips detection is described. As part of the objectives, we also try to acquire information to represent the lips. GA is very useful for optimization of parameters in complex image matching. When GA is applied to the real-time processing, the search speed is a issue with keeping the accuracy. Using small population is one of the approaches to improve the GA search speed. However, this can cause the low accuracy, because diversity of the population will be lost. In order to avoid this problem, we propose a downsized GA with automatic search domain control. In this paper, the optimization problem is template matching. The target image includes a face of talking person and has significant changes of a whole scene by camera motion. The template image is only one closed mouth and prepared for each environment and person, because of personal use. In simulations, a simple image is used, which is the basis of video sequence. Our proposed method is verified by comparison of a process of evolution between a downsized GA and a standard GA.
255.
Fukuda Keiji, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Personal Authentication using Fingerprints, Proc. of the International Conference on Soft Computing and Intelligent Systems and International Symposium on advanced Intelligent Systems, 1144-1149, Tokyo, Sep. 2006.
256.
Takizawa Atsushi, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Recognizing Parked Vehicles License Plate Using Image Frame-Difference and Template Matching, Proc. of the International Conference on Soft Computing and Intelligent Systems and International Symposium on advanced Intelligent Systems, 141-144, Tokyo, Sep. 2006.
(Summary)
In this paper, parked vehicles license plate recognition using image frame-difference and template matching is proposed. Image frame-difference is used to extract the vehicle area, reducing the license plate search domain, thereby reducing the total search time. The processes of expansion and reduction are then applied to reduce the noise in the extracted plate domain. Based on the data from 83 vehicles, the most likely plate area is then decided. License plate recognition is carried out using template matching. For each possible character, three templates are prepared. We then performed computer simulations to show the effectiveness of our method. From the result, it was confirmed that the preprocessing method was effective.
257.
Yuji Matsumura, Minoru Fukumi and Yasue Mitsukura : Hybrid EMG Recognition System by MDA and PCA, Proc. of 2006 International Joint Conference on Neural Networks (WCCI'2006), 10750-10756, Vancouver, Jul. 2006.
(Summary)
In this paper, we propose a recognition system of wrist operation by focusing on ElectroMyoGram (EMG), that is, the living body signal generated with movement of a subject. In previous research, we only performed pattern recognition by Neural Network (NN) and Fast Fourier Transform (FFT). In contrast, in proposal research, we try to improve recognition accuracy and reduce learning-time of system by combining Multi Discriminant Analysis (MDA) and gradual Principal Component Analysis (PCA) based on the PCA result of EMG data. From results of computer simulation, it is shown that our approach is effective for improvement in recognition accuracy and speed. This system is partly implemented on the DPS learning board.
258.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru and Minoru Fukumi : Eye Detection and Tracking using Genetic Algorithm, Proceeding of the 4th International Conference on Computing, Communications and Control Technologies, 26-31, Orlando, Jul. 2006.
259.
Stephen Githinji Karungaru, Minoru Fukumi, Norio Akamatsu and Akashi Takuya : Optimizing Feature Extraction for the Camera Mouse using Genetic Algorithms, WSEAS Proceeding on Computers, 873-877, Jul. 2006.
260.
Tadahiro Oyama, Yuji Matsumura, Stephen Githinji Karungaru, Yasue Mitsukura and Minoru Fukumi : Construction of Wrist Motion recognition System, Proc. of 2006 RISP International Workshop on Nonlinear Circuits and Signal Processing, 385-388, Honolulu, Mar. 2006.
(Summary)
Recently, studies of artificial arms and pointing devices using ElectroMyoGram(EMG) have been actively done. However, the individual variation of EMG is large, and its repeatability is low. Furthermore, EMG is usually measured from a part with comparatively big muscular fibers such as arms and shoulders. Therefore, if we can recognize wrist operations using EMG which was measured from the wrist, the range of application will extend furthermore. In this study, we aim toward the development of a device of wristwatch type that consolidates operational interface of various equipments. In particular, as an early stage, we propose a wrist motion recognition system. First, we execute the Fourier transform to the signal for feature extraction. Next, we experiment it by using neural networks after the dimensional reduction by using Simple-PCA and Simple-FLDA to reduce the number of inputs to NN. It was confirmed that it was one of the techniques which were effective in the wrist recognition experiment.
261.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Method of Gender and Age Estimation Based on Facial Knowledge, Proc. of 2006 RISP International Workshop on Nonlinear Circuits and Signal Processing, 373-376, Honolulu, Mar. 2006.
(Summary)
The purpose of this paper is to propose a method of gender and age estimation which is robust for environmental changing. We propose a feature-point detection method which is the Advanced Retinal Sampling Method (ARSM), and a feature extraction method. As features for the gender and age estimation, facial shape, skin texture, hue and Gabor-feature are used. We examined the left-right symmetric property of the face concerning gender and age estimation by the proposed method.
262.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Graininess Suppression Method for Image Restoration Based on Saving Edge Shape, Proc. of 2006 RISP International Workshop on Nonlinear Circuits and Signal Processing, 235-238, Honolulu, Mar. 2006.
(Summary)
Previous studies of image restoration for noise image were based on a mask processing. These conventional noise removal methods expressed from the mask processing have an issue of defining degradation to accompany spacial processing. In this paper, we propose a graininess suppression method based on edge shape. In this method, we detect edges from an noisy image and perform graininess suppression for this image based on edge information. On the edge detection, we execute an image transformation for an image that enables us to extract edge by making a principal component image. Moreover, we use the canny edge detection operator, which can detect a weak edge that relates to a real edge, and do not detect a lie edge. In the suppression process, we use the Wiener filter that can restore an noisy image without making a complete edge map and the original signal map. We demonstrated the effectivness of the present method for the noise added image by means of computer simulations.
263.
Yasue Mitsukura, Kensuke Mitsukura and Minoru Fukumi : Face Detection from Near-Infrared Camera Images Using the GA and NN, Proc. of 2006 RISP International Workshop on Nonlinear Circuits and Signal Processing, 219-222, Honolulu, Mar. 2006.
(Summary)
The purpose of this research is to recognize a face with an near-infrared camera. The face detection that used images from near-infrared camera is comparatively difficult to be done, because they are gray scale images. In this paper, the filter by using GA is designed, and the method of detecting the face and the position from the near-infrared images is proposed. It is demonstratet that our approach is effective for vehicle driver monitoring.
264.
Satomi Ota, Yasue Mitsukura and Minoru Fukumi : A Learning Method for Making the EEG Maps, Proc. of 2006 RISP International Workshop on Nonlinear Circuits and Signal Processing, 187-190, Honolulu, Mar. 2006.
(Summary)
Recently in the world, the diseases based on the stress are increasing. It is thought that the living in a comfortable space is important to prevent the diseases beforehand. A feeling of comfortable depends on a person, therefore to consider the personality is necessary beforehand. Then, we research the relation between the external stimulation and human by using the EEG as an index. In this study, our viewpoint is to detect the effectiveness of the music relaxation. Therefore, in order to make the ``personal comfortable'', we show the external stimulation examples by the subjects and some genre music. First, we measure the EEG of subjects in listening to each genre of music in our approach. Next, we extract some features of EEG patterns by using the neural networks. Finally, we make the EEG map displaying the features data in two dimensions to clarify the location of each genre in the EEG.
265.
Takuya Akashi, Yuji Wakasa, Kanya Tanaka, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Genetic Eye Detection System, Proc. of 2006 RISP International Workshop on Nonlinear Circuits and Signal Processing, 377-380, Honolulu, Mar. 2006.
(Summary)
In this paper, high-speed size and orientation invariant eye tracking in an active scene using template matching and genetic algorithms is proposed. As part of the objectives, we also try to acquire numerical parameters to represent the eye, for example, position, scaling, and rotation angle. The information is useful for many applications, where high performance is required, such as eye gaze detection or estimation for robot perception and mobile devices interfaces. The difficulty in eye tracking is mainly due to motion of the human head and the active scene by free camera motion. Our proposed system is based on template matching using genetic algorithms (GAs). Only one artificial template is used for every human user. Various geometric changes of iris can be supported by genetic algorithms. In this paper, eye tracking system with genetic algorithms is proposed, and selection of the best template image is described. We achieved a eye detection accuracy of 74 % at an average processing time of 0.03 seconds.
266.
Keiko Sato, Yasue Mitsukura and Minoru Fukumi : A Kansei Extraction Method from the Individual Characteristics by Using the Color Chart, Proc. of 2006 RISP International Workshop on Nonlinear Circuits and Signal Processing, 142-145, Honolulu, Mar. 2006.
(Summary)
The color design is one of the important elements that influence the impression of products, therefore the technology which understands and reflects the consumer's sensibility is needed in the color design. Especially, the color coordination system that connects colors with impressions is expected as the system supporting the color design. Therefore, in the field of the KANSEI ( it is named by Japanese) engineering, researches that model the correlation of colors (the arrangement of the color) and impression words have been done. However, because the impression received from the color is various by individual, it is necessary to model the KANSEI information to each individual. In this paper, we analyze the data of the evaluation questionnaire of the color for the subject. Then, we propose the KANSEI extraction method from the subject informations. Concretely, by using the principal component analysis (PCA), the correlation with an amount of the sensibility and the color feature is found. As a result, it is thought that the KANSEI model of the subject can be made, and the feature of the preference can be analyzed.
267.
Michiyo Nishioka, Yasue Mitsukura, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Measurement of Skin Texture Using Artificial Neural Network and SOM, Proc. of 2006 RISP International Workshop on Nonlinear Circuits and Signal Processing, 33-36, Honolulu, Mar. 2006.
(Summary)
In this paper, the purpose is to offer a different diagnostic method for skin texture evaluation using sophisticated skin diagnostic system that a layperson can easily use at home. The experiments for evaluation were carried out using an artificial neural network (NN) and a Self-Organizing Map (SOM) to measure the degree of similarity. The Fast Fourier Transform (FFT) was performed to skin images and the number of skin textures counted used as the input data. Then we use Simple Principal Component Analysis (SPCA) in order for PCA to achieve high speed processing. In this experiment, we use newly SPCA and could obtain effective result comparison with before. Moreover, the mapping images of SOM and SPCA were obtained the result that is nearby positioned the similar image of skin.
268.
Stephen Githinji Karungaru, Youngmin Choie, Minoru Fukumi and Norio Akamatsu : Korean Lightweight Font Development using Outline Compression Reduction Function, Proc. of The 12th Korea-Japan Joint Workshop on Frontiers of Computer Vision, 289-294, Tokushima, Feb. 2006.
269.
Yoshiki Kubota, Yasue Mitsukura, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Automatic Extraction System of a Kidney Region from CT Images, CD Proc. of 9th International Conference on Mechtornics Technology, Vol.1, No.T7-3, 1-5, Kuala Lumpur, Dec. 2005.
(Summary)
In this paper, a kidney region extraction method as a preprocessing of kidney disease detection is proposed. The kidney region is detected based on its contour information that is extracted from a CT image using a dynamic gray scale value refinement method based on the Q-learning. An initial point to extract the kidney contour is decided by training gray scale values along horizontal direction using a Neural Network (NN). Furthermore the kidney contour is corrected by using SNAKES. It is demonstrated that the proposed method can accurately detect the kidney contour from CT images of any patient.
270.
Takahiro Ogawa, Yasue Mitsukura, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Relevance of the EEG and feeling during listening to the music, CD Proc. of 9th International Conference on Mechtornics Technology, Vol.1, No.T7-5, 1-5, Kuala Lumpur, Dec. 2005.
(Summary)
In order to solve stress problems, researchers have studied music therapy. In this paper, the purpose is extraction of features that may be influenced by the music. We pay attention to EEG (electroencephalogram) as an objective and absolute scale. This paper proposes a method that extracts features of the EEG by PCA (principal component analysis) and CDA (canonical discriminant analysis). Then we analyze each feature data by NN (neural network). In order to examine whether the proposal system is effective, we try computer simulations for the EEG classification. According to recognition rate by the NN, it was considered that the CDA extracted and classified the features of the EEG better than the PCA.
271.
Ryohei Haga, Yasue Mitsukura, Stephen Githinji Karungaru, Minoru Fukumi, Norio Akamatsu and Motokatsu Yasutomo : Automatic Extraction of Left Ventricle by Using X-Ray Photogram, CD Proc. of 9th International Conference on Mechtornics Technology, Vol.1, No.T2-3, 1-5, Kuala Lumpur, Dec. 2005.
(Summary)
Recently, as sophisticated medical instruments are being developed, the inside state of a human body has become well known. However, doctor's burden becomes heavier because the number of images that are taken with medical instruments per person drastically increases. In such cases, the development of automatic extraction system for the organs is needed. In this paper, we propose the extraction the left ventricle in such images using the Gabor transform and SNAKES. The kind of image processed in this work is the X-ray photograms of the left ventricle by cardiac catheterization.
272.
Yoshiyuki Morioka, Yasue Mitsukura, Stephen Githinji Karungaru, Minoru Fukumi, Norio Akamatsu and Motokatsu Yasutomo : Contour Extraction of Heart from Ultrasonic Image Using SNAKES, CD Proc. of 9th International Conference on Mechtornics Technology, Vol.1, No.T7-3, 1-4, Kuala Lumpur, Dec. 2005.
(Summary)
Recently, as sophisticated medical instruments have been developed, the inside state of a human body has become well-known. In particular, in the examination of heart, non-surgical and non-intrusion ultrasonic diagnostics are generally utilized. However, the diagnosis, depends on the doctor's subjectivity and experience. Furthermore, the doctor's burden increases because the contours of the heart are extracted by manual operations. Therefore, the development of automatic diagnostic imaging systems is needed. In this paper, the contour of the entire heart and left ventricle is extracted by using an active contour model SNAKES as a preprocessing of the heart disease detection. The entire heart contour made with SNAKES is used as a control point in the next image to automate contour tracking.
273.
Yuji Matsumura, Minoru Fukumi and Stephen Githinji Karungaru : Wrist Motion Recognition System by Neural Networks, CD Proc. of 9th International Conference on Mechtornics Technology, Vol.1, No.T6-3, 1-6, Kuala Lumpur, Dec. 2005.
(Summary)
We propose a recognition system based on wrist movements by focusing on ElectroMyoGram (EMG), that is, the living body signal generated with voluntary movement of subject muscles as the initial stage for construction of the total operation device. In this paper, we aim for construction of a high-speed and high-accurate EMG recognition system using Fast Fourier Transform (FFT) for feature extraction, Simple-PCA (SPCA) for feature compression, and Neural Networks (NN) for recognition. Furthermore, we select evaluation data using Self-Organizing Map (SOM) for recognition accuracy improvement. From results of computer simulation, it is shown that our approach is effective in the improvement of recognition accuracy.
274.
Minoru Fukumi, Yohei Takeuchi, Yasue Mitsukura, Hiroko Fukumoto and Khalid Marzuki : Neural Network Based Threshold Determination for Malaysia License Plate Character Recognition, Proc. of 9th International Conference on Mechtornics echnology, Vol.1, No.No.T1-4, 1-5, Kuala Lumpur, Dec. 2005.
(Summary)
In this paper, a method to recognize characters of vehicle license plate in Malaysia by using a neural network based threshold method is presented. Vehicle license plate recognition is one of important techniques that can be used for the identification of vehicles all over the world. There are many applications such as entrance admission, security, parking control, airport or harbor cargo control, road traffic control, speed control, toll gate automation and so on. For separation of characters and background, a threshold of digitalization is important and is determined using a three-layered neural network in this paper. Furthermore, in the extracted character portions, we segment characters and recognize them by obtaining their features. Computer simulations show that character segmentation and recognition can be effectively carried out using the present method.
275.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : License Plate Localization Using Template Matching, CD Proc. of 9th International Conference on Mechtornics Technology, Vol.1, No.T4-3, 1-5, Kuala lumpur, Dec. 2005.
276.
kensuke Mitsukura, Minoru Fukumi, Yasue Mitsukura and Motokatsu Yasutomo : Detectioning of an Asynergy Using the Neural Network, Proc. of International Conference on Computational Intelligence for Modelling, Control and Automation, Vol.1, No.2, 87-91, Wien, Nov. 2005.
(Summary)
Recently in medical fields, various imaging diagnostic technologies have been studied and used in practical. It is necessary to develop an automatic diagnosing processing system for detecting and diagnosing the internal organs. By the way, cardiac disease is one of the most common cause of death. Therefore, it is necessary to measure cardiac function quantitatively. The processing images are X-ray photograms of the left ventricle by cardiac catheterization. In this paper, we propose the detection system of asynergy in the left ventricle by using a neural network. Furthermore, in order to demonstrate the effectiveness of the proposed method, we show the simulation example by using the real data
277.
Minoru Fukumi, Stephen Githinji Karungaru and Yasue Mitsukura : Feature Generation by Simple-FLDA for Pattern Recognition, Proc. of International Conference on Computational Intelligence for Modelling, Control and Automation, Vol.1, No.2, 730-734, Wien, Nov. 2005.
(Summary)
In this paper, a new feature generation method for pattern recognition is proposed, which is approximately derived from geometrical interpretation of the Fisher linear discriminant analysis (FLDA). Recently the FLDA has been used in many fields, especially face image analysis. The drawback of FLDA is a long computational time based on a large-sized covariance matrix and the issue that the within-class covariance matrix is usually singular. Generally FLDA has to carry out minimization of a within-class variance. However in this case the inverse matrix of the within-class covariance matrix cannot be obtained, since data dimension is generally higher than the number of data and then it includes many zero eigenvalues. In order to overcome this difficulty, a new iterative feature generation method, a simple FLDA is introduced and its effectiveness is demonstrated for pattern recognition problems.
278.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Malaysian License Plate Recognition Using Artificial Neural Networks and Evolutionally Computation, Proc. of International Federation for Systems Research, Vol.1, No.S3-7-3, 1-5, Kobe, Nov. 2005.
279.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Automatic Facial Gesture Construction Using Image Warping, Proc. of International Federation for Systems Research, Vol.1, No.S3-2-2, 1-6, Kobe, Nov. 2005.
280.
Miyoko Nakano and Minoru Fukumi : Age and Gender Recognition Using Facial Edge Information, Proc. of The First World Congress of the International Federation for Systems Research, Vol.1, 206-208, Kobe, Nov. 2005.
(Summary)
For age and gender recognition, this study focuses on the edge information that consists of all wrinkles in a face and also a neck. Density histogram in which the value of the edge intensity of the vertical direction and the horizontal direction in an image is greater than a threshold value is computed by using the edge feature extraction. They are treated as input data to a neural network. These simulations are tried toward the face image database which collected face images of 15-64 years old. We perform the psychology experiment to know how humans are recognizing human's age. The result is compared with the result by the neural network. In order to show the effectiveness of the proposed method, computer simulations are carried out using these face images.
281.
Miyoko Nakano and Minoru Fukumi : Age and Gender Classification by Using Edge Feature Extraction, Proc. of International Conference on Neural Information Processing, Vol.1, 355-359, Taipei, Nov. 2005.
(Summary)
For age and gender classification, this study focuses on the edge information that consists of all wrinkles in a face and also a neck. Density histogram in which the value of the edge intensity of the vertical direction and the horizontal direction in an image is greater than a threshold value is computed by using the edge feature extraction. They are treated as input data to a neural network. This classification is tried toward the face image database which collected face images of 15-64years old. In order to show the effectiveness of the proposed method, computer simulations are carried out using these face images.
282.
Minoru Fukumi and Yasue Mitsukura : A Simple Feature Generation Method Suitable for Pattern Recognition, Proc. of International Conference on Neural Information Processing, Vol.1, 576-580, Taipei, Nov. 2005.
(Summary)
-This paper presents a new feature generation method suitable for pattern recognition, which is approximately derived from geometrical interpretation of the Fisher linear discriminant analysis (FLDA). Recently the FLDA has been used in many fields, especially face image analysis. The drawback of FLDA is a long computational time based on a large-sized covariance matrix and the issue that the within-class covariance matrix is usually singular. Generally FLDA has to carry out minimization of a within-class variance. However in this case the inverse matrix of the within-class covariance matrix cannot be obtained, since data dimension is generally higher than the number of data and then it includes many zero eigenvalues. In order to overcome this difficulty, a new iterative feature generation method, a simple FLDA is introduced and its effectiveness is demonstrated.
283.
Minoru Fukumi and Yasue Mitsukura : Feature Generation by Simple FLD, Proc. of 9th International Conference on Knowledge-Based Intelligent Information & Engineering Systems, Vol.1, 643-649, Melbourne, Sep. 2005.
(Summary)
This paper presents a new algorithm for feature generation, which is approximately derived based on geometrical interpretation of the Fisher linear discriminant analysis. Recently the Fisher linear discriminant (FLD) analysis has been used in such a field, especially face image analysis. The drawback of FLD is a long computational time in compression of large-sized between-class and within-class covariance matrices. Usually FLD has to carry out minimization of a within-class variance. However in this case the inverse matrix of the within-class covariance matrix cannot be obtained, since data dimension is higher than the number of data and then it includes many zero eigenvalues. In order to overcome this difficulty, a new iterative feature generation method, a simple FLD is introduced and its effectiveness is demonstrated.
284.
Minoru Fukumi, Taketsugu Nagao, Yasue Mitsukura and Rajiv Khosla : Drift Ice Detection Using a Self-Organizing Neural Network, Proc. of 9th International Conference on Knowledge-Based Intelligent Information & Engineering Systems, Vol.1, 1268-1274, Melbourne, Sep. 2005.
(Summary)
This paper proposes a segmentation method of SAR (Synthetic Aperture Radar) images based on a SOM (Self-Organizing Map) neural network. SAR images are obtained by observation using microwave sensor. For teacher data generation, they are segmented into the drift ice (thick and thin), and sea regions manually, and then their features are extracted from partitioned data. However they are not necessarily effective for neural network learning because they might include incorrectly segmented data. Therefore, in particular, a multi-step SOM is used as a learning method to improve reliability of teacher data, and carry out classification. This process enable us to fix all mistook data and segment the SAR image data using just data. The validity of this method was demonstrated by means of computer simulations using the actual SAR images.
285.
Keiko Sato, Yasue Mitsukura and Minoru Fukumi : Emotional Extraction System by Using the Color Combination, Knowledge-Based Intelligent Infomation and Engineering Systems, Melbourne, Sep. 2005.
286.
Yoshiki Kubota, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Motokatsu Yasutomo : Automatic Extraction System of a Kidney Region Based on the Q-Learning, Knowledge-Based Intelligent Infomation and Engineering Systems, Melbourne, Sep. 2005.
(Summary)
In this paper, a kidney region is extracted as a preprocessing of kidney disease detection. The kidney region is detected based on its contour information that is extracted from a CT image using a dynamic gray scale value refinement method based on the Q-learning. An initial point to extract the kidney contour is decided by training gray scale values along horizontal direction with Neural Network (NN). Furthermore the kidney contour is corrected by using the snakes more accurately. It is demonstrated that the proposed method can detect stably the kidney contour from CT images of any patients.
287.
Hideyuki Takajo, Minoru Fukumi and Norio Akamatsu : Prediction of Foul Ball Falling Spot in a Base Ball Game, Knowledge-Based Intelligent Infomation and Engineering Systems, Melbourne, Sep. 2005.
288.
Takahiro Ogawa, Satomi Ota, Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Influence of Music Listening on the Cerebral Activity by Analyzing EEG, Knowledge-Based Intelligent Infomation and Engineering Systems, Melbourne, Sep. 2005.
289.
Minoru Fukumi : Driver Face Monitoring Using a Near-Infrared Camera, Proc. of Signal and Information Processing'2005, 156-160, Honolulu, Aug. 2005.
(Summary)
This paper presents a method for driver face monitoring which includes face detection and estimation of its direction. In particular, the present method tries to extract a face from gray scale images obtained using a near-infrared camera, which is equipped in the top of the automobile windshield. This method is based on a template matching (SSDA method) with multi templates and a neural network (NN) for face detection. After template matching, NN tries to recognize a face and its direction. Furthermore, the simple PCA is used to yield features suitable for face detection. From results of computer simulations, it is shown that our approach can give a good result for vehicle driver face monitoring.
290.
Minoru Fukumi and Yasue Mitsukura : A Simple Feature Generation Method Based on Fisher Linear Discriminant Analysis, Proc. of Signal and Information Processing'2005, 342-346, Honolulu, Aug. 2005.
(Summary)
This paper presents a new iterative algorithm for feature generation, which is approximately derived based on geometrical interpretation of the Fisher linear discriminant (FLD) analysis. The drawback of FLD is a long computational time in compression of a large-sized between-class covariance and within-class covariance matrices. Usually FLD has to carry out minimization of a within-class variance. However in this case the inverse matrix of the within-class covariance matrix cannot be obtained. In order to overcome this difficulty, a new iterative feature generation method, a simple FLD is introduced and its effectiveness is demonstrated.
291.
Takahiro Ogawa, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Classification of Electroencephalogram in Listening to the Music by Multivariate Analysis, SICE Annual Conference, Okayama, Aug. 2005.
292.
Satomi Ota, Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : The EEG Feature Extraction Using the Principal Component Analysis, SICE Annual Conference, Okayama, Aug. 2005.
293.
Keiko Sato, Yasue Mitsukura and Minoru Fukumi : Designing the KANSEI information System Using the Color Coordination, SICE Annual Conference, Okayama, Aug. 2005.
(Summary)
Recently, many researches using the human interface have been done. In particular, the KANSEI information processing is attracted as the multimedia information processing on the human interface. The color coordination system which connects colors with feelings is expected as the system supporting the color design. Therefore, to analyze the relation between colors and feelings is one of problems in the field of the KANSEI engineering. In this research, the method for judging the impression caused by the color automatically is proposed. In this paper, the correlation with the impression caused by the color and the color feature is analyzed as the first stage of this research. Concretely, by using the principal component analysis, the correlation with an amount of the sensibility and the color feature is found.
294.
Yoshiki Kubota, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Automatic Extraction of a Kidney Contour by Narrowing Region Based on the Q-Learning, SICE Annual Conference, Okayama, Aug. 2005.
(Summary)
At present, owing to an aging population and Western-style food, the number of kidney disease patients in Japan is increasing. It is difficult for people to recover fully from kidney disease. Early detection of a kidney disease is therefore needed. But diagnosis based on CT images has faults that are time-consuming and a great labor is required since the quantity of CT images is huge. We propose a method that automatically extracts the kidney region as a preprocessing of kidney failure detection. The kidney region is detected based on contour information that is extracted from the CT image using a dynamic gray scale value refinement method by Q-learning. It is demonstrated that the proposed method can stably detect the kidney from CT images of any patients.
295.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Face Gesture Recognition Using Artificial Neural Networks, Proc. of The Society of Instrument and Control Engineers (SICE) Annual Conference, 136-140, Okayama, Aug. 2005.
296.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Simple Algorithm of Pitch Detection by using Fast Direct Transform, Proceedings 2005 IEEE International Symposium on Computational Intelligence in Robotics and Automation, Finland, Jun. 2005.
(Summary)
There are some fundamental frequency detection methods for speech processing such as cepstrum analysis. However, the traditional methods need a lot of computing time and arithmetic processing. Moreover, a speech signal is converted into frequency domain by using an analysis section. In this paper, we propose a fast direct transformation (FDT). FDT extracts an amplitude feature of a signal by a simple computation. We perform the fundamental frequency detection of speech signal by using FDT. We compare the FDT algorithm with the conventional fundamental frequency detection methods by using teacher data (fundamental frequency) detected by the inspection. We perform an improvement as compared with an autocorrelation method by using FDT algorithm.
Miyoko Nakano, Fumiko Yasukata and Minoru Fukumi : Marketing Data Collection from Face Images Using Neural Networks, Proc. of IEEE International Workshop on Nonlinear Signal and Image Processing, 212-216, Sapporo, May 2005.
(Summary)
In the marketing field, it is very useful to collect the customer data of age and gender etc. For gender and age classification, this study focuses on the edge information that consists of all wrinkles in a face and also a neck. A density histogram in which the value of the edge intensity of the vertical direction and the horizontal direction in an image is greater than a threshold value is computed by using the edge information. They are treated as input data to a neural network. This classification is used on a face image database in which are the collected face images of people 15-64 years old. In order to show the effectiveness of the proposed method, computer simulations are carried out using these face images.
298.
Yuji Matsumura, Minoru Fukumi and Norio Akamatsu : Wrist Motion Recognition System by EMG, Proc. of the First International Conference on Complex Medical Engineering, 216-221, Kagawa, May 2005.
(Summary)
We propose a recognition system based on wrist movements by focusing on ElectroMyoGram (EMG), that is, the living body signal gener-ated with voluntary movement of subject muscles as the initial stage for construction of the total operation device. In this pa-per, we aim for construction of a high-speed and high-accurate EMG recognition system using Fast Fourier Transform (FFT) for feature extraction, Simple-PCA (SPCA) for feature com-pression, and Neural Networks (NN) for recognition. Further-more, we reduce the noise using Wavelet Transform for practi-cal application. From results of computer simulation, it is shown that our approach is effective for improvement in rec-ognition accuracy and speed.
299.
Michiyo Nishioka, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Measurement of Skin Texture Using Genetic Programming and Neural Network, Proc. of the First International Conference on Complex Medical Engineering, 740-744, Kagawa, May 2005.
(Summary)
In this paper, the purpose of this experiment is to offer a different diagnostic method using sophisticated skin diagnostic system which a layman can easily use at home. A binarization method of images for presumption of skin age is proposed using the genetic programming (GP) which is an evolutional computing method. Skin age can be estimated by skin texture, but it is difficult to binarize skin images owing to individual differences. Therefore, results obtained by a genetic image processing were achieved with consideration of individual differences, and good results were obtained. Furthermore another experiment was carried out using neural network (NN). The fast fourier transform (FFT) was performed to skin images and the classification experiment was carried out by a neural network. Then we began to use newly Simple Principal Component Analysis (SPCA) in order for PCA to achive high speed processing. However it is demonstrated that an effective classification result was
300.
Ryouhei Haga, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Automatic Detection of Left Ventricular Asynergy by Fuzzy Reasoning, Proc. of the First International Conference on Complex Medical Engineering, 826-831, Kagawa, May 2005.
(Summary)
As Japanese daily life is Americanized, heart diseases such as angina and myocardial infarction are increasing. We need to observe consecutive cardiac muscle motion to detect their diseases. In this paper the left ventricular axis and the contact points in the heart region are defined, and then cardiac muscle momentum is extracted. We discriminate an abnormal case and a normal case by using a neural network and fuzzy reasoning to confirm the effectiveness of our approach. Finally, in order to show the effectiveness of the proposed method, we show the simulation examples by using real images.
301.
Yoshiki Kubota, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Automatic Extraction of a Kidney Region by Using the Q-learning, Proc. of the First International Conference on Complex Medical Engineering, 832-836, Kagawa, May 2005.
(Summary)
In this paper, a kidney region is extracted as a preprocessing of kidney disease detection. The kidney region is detected based on its contour information that is extracted from a CT image using a dynamic gray scale value refinement method based on the Q-learning. An initial point to extract the kidney contour is decided by training gray scale values along horizontal direction with Neural Network (NN). Furthermore the kidney contour is corrected by using the snakes more accurately. It is demonstrated that the proposed method can detect stably the kidney contour from CT images of any patients.
302.
Takahiro Ogawa, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Discovery of Influence in Music Listening by Analysing EEG, Proc. of the First International Conference on Complex Medical Engineering, 691-694, Kagawa, May 2005.
(Summary)
In this paper, the purpose is extraction of features that may be influenced by the music. We pay attention to EEG (electroencephalogram) as an objective and absolute scale. In this paper, we propose a method that extracts features of the EEG by PCA (principal component analysis) and CDA (canonical discriminant analysis). Then we analyze each feature data by NN (neural network). In order to examine whether the proposal system is effective, we try computer simulations for the EEG classification. According to recognition rate by the NN, it was considered that the CDA extracted and classified the features of the EEG better than the PCA.
303.
Satomi Ota, Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Feature Extraction of EEG Patterns Using the Principal Component Analysis and Neural Networks, Proc. of 2005 RISP International Workshop on Nonlinear Circuits and Signal Processing, 319-321, Honolulu, Mar. 2005.
(Summary)
In this paper, a system automatically selects the suitable music for music therapy from a large database of music by each EEG. By using this system, music therapy is done regardless of time and place, and it is needed for constructing this system to clear relations between the music and the EEG. In this paper, we measure the EEG of subjects in listening to music and extract some features of their patterns by using the Principal Component Analysis (PCA). Then we analyze them by using the neural networks (NN). Finally, in order to demonstrate the effective of the proposed method, we carry out the computer simulation. Then, we show the effectiveness of the proposed method.
304.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Face Detection: Size and Rotation Invariance using Genetic Algorithms, Proc. of 2005 RISP International Workshop on Nonlinear Circuits and Signal Processing, 211-214, Honolulu, Mar. 2005.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Automatic Face Metamorphosis in Color Images, Proc. of 2005 RISP International Workshop on Nonlinear Circuits and Signal Processing, 131-134, Honolulu, Mar. 2005.
Keioko Sato, Mihoko Kimura, Yasue Mitsukura and Minoru Fukumi : The Feeling Classification by the Color based on the KANSEI Information Processing, Proc. of 2005 RISP International Workshop on Nonlinear Circuits and Signal Processing, 275-277, Honolulu, Mar. 2005.
(Summary)
In this research, the method for judging the image caused by the color automatically is proposed. In this paper, the correlation with the image caused by the color and the color feature is analyzed as the first stage of this reserch. Concretely, by using the principal component analysis, the correlation with an amount of the sensibility and the color feature is found.
307.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Pitch Detection for Speech Processing by Using a Fast Direct Transformation, Proc. of 2005 RISP International Workshop on Nonlinear Circuits and Signal Processing, 351-354, Honolulu, Mar. 2005.
(Summary)
There are some pitch detection methods for speech processing such as cepstrum analysis. However, the traditional methods need a lot of computing time and arithmetic processing. Moreover, a speech signal is converted into frequency domain by using an analysis section. In this paper, we propose a fast direct transformation (FDT). FDT extracts an amplitude feature of a signal by a simple computation. We perform the pitch detection of speech signal by using FDT. We compare the FDT algorithm with the conventional pitch detection methods by using teacher data(pitch) detected by the inspection. It is shown that the improvement rate in extraction of fundamental frequency is 21.3%.
308.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Three-Dimensional Geometric Information Acquisition of Lips for Natural Scene, Proc. of 2005 RISP International Workshop on Nonlinear Circuits and Signal Processing, 123-126, Honolulu, Mar. 2005.
(Summary)
We describe a lips region detection system of a talking person in natural scenes. In particular, we try to acquire numerical parameters to represent the lips information, during speech. Therefore, this lips region extraction must be robust for some considerable geometric changes, as well as varying lips shapes by speech. The present method is image template matching with genetic algorithms (GAs) and projective geometry. We use only one camera. Input images are a template and a target image. The number of template images is only one and it is prepared for each subject. The target image contains the face and the background.
309.
Takuya Akashi, Hironori Nagayama, Minoru Fukumi and Norio Akamatsu : Estimation of Face Direction Using Near-Infrared Camera, Proc. of 2005 RISP International Workshop on Nonlinear Circuits and Signal Processing, 231-234, Honolulu, Mar. 2005.
(Summary)
This paper presents a method for human face detection and estimating its direction. In particular, the present method tries to extract a face from gray scale images obtained using a near-infrared camera, which is equipped in the top of the automobile windshield. This method is based on template matching (SSDA method) with multi templates and a neural network (NN). After the template matching, NN tries to recognize a face and its direction. Furthermore, the simple PCA is used to yield features suitable for face detection. From results of computer simulations, it is shown that our approach can give a good result compared to another approach.
310.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Real-Time Genetic Lips Region Detection and Tracking in Natural Video Scenes, Proc. of IEEE Conference on Cybernetics and Intelligent Systems'2004, 682-687, Singapore, Dec. 2004.
(Summary)
In this paper, real-time detection and tracking of lips region of a talking person in natural scenes is addressed. In particular, we try to acquire numerical parameters to represent the lips information. Because, this information is very important for many applications, such as audio-visual speech recognition, robot perception, and interface of mobile devices. The difficulty lies in deformations and geometric change of lips, by speech and free camera work. Our proposed system is based on template matching with genetic algorithms (GAs). In our previous system, there is a trade-off between accuracy and a processing time. However, we can overcome this by two new methods: (a) a flexible control of a search domain, (b) inheritance of genetic information between video frames. We demonstrated the effectiveness of our proposed system by using some 5 seconds video sequences. The average results are that the accuracy is 94,44% and the processing time is 4.50 seconds.
Toshiki Kubota, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Automatic Extraction of a Kidney Region by Using the Q-learning, Proc. of International Symposium on Intelligent Signal Processing and communication Systems'2004, P007-1-P007-5, Seoul, Nov. 2004.
(Summary)
At present, owing to an aging population and Western-style food, the number of kidney disease patients in Japan is increasing. It is difficult for people to recover fully from kidney disease. Early detection of a kidney disease is therefore needed. But diagnosis based on CT images has faults that are time-consuming and a great labor is required since the quantity of CT images is huge. We propose a method that automatically extracts the kidney region as a preprocessing of kidney failure detection. The kidney region is detected based on contour information that is extracted from the CT image using a dynamic gray scale value refinement method by Q-learning. It is demonstrated that the proposed method can stably detect the kidney from CT images of any patients.
312.
Michiyo Nishioka, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Measurement of Skin texture Using Genetic Image Analysis, Proc. of International Symposium on Intelligent Signal Processing and communication Systems'2004, P060-1-P060-5, Seoul, Nov. 2004.
(Summary)
A binarization method of images for presumption of skin age is proposed using genetic programming (GP) which is an evolutionary computing method. Skin age can be estimated by skin texture, but it is difficult to binarize skin images owing to individual differences. Therefore, results obtained by genetic image processing were achieved with consideration of individual differences, and good results were obtained. The fast Fourier transform (FFT) was performed on skin images and a classification experiment was carried out by a neural network (NN) as a comparative experiment. However it is demonstrated that an effective classification result was not obtained by FFT and NN compared to our approach.
313.
Tomohiko Nukano, Minoru Fukumi and Marzuki Khalid : Vehicle License Plate Character Recognition by Neural Networks, Proc. of International Symposium on Intelligent Signal Processing and communication Systems'2004, P056-1-P056-5, Seoul, Nov. 2004.
(Summary)
This paper tries to recognize characters of vehicle license plates in Malaysia. Vehicle license plate recognition is one of important techniques that can be used for the identification of vehicles. It is useful in many applications such as entrance admission, security, parking control, airport or harbor cargo control, road traffic control, speed control, toll gate automation and so on. In this paper, when part of character and background are dissociated, the threshold of digitalization is important and is determined using a neural network with a three-layered structure. Furthermore, in the extracted character portions, we recognize them by obtaining the features of directional line-elements. Computer simulations show that character recognition can be effectively carried out using such a method.
314.
Ryouhei Haga, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Motokatsu Yasutomo : Automatic Detection of Left Ventricular Asynergy by Fuzzy Reasoning, Proc. of International Symposium on Intelligent Signal Processing and communication Systems'2004, S15-3-1-S15-3-5, Seoul, Nov. 2004.
(Summary)
Recently, as sophisticated medical instruments have been developed, the internal state of a human body has become well-known. However, a doctor's burden becomes heavier because the number of images which are taken per person with medical instruments drastically increases. Therefore, the development of automatic diagnostic imaging systems is needed. Incidentally, as Japanese daily life is Americanized, heart diseases, such as angina and myocardial infarction, are increasing. We need to observe consecutive cardiac muscle motion to detect their diseases. The left ventricular axis and the contact points in the heart region are defined, and then cardiac muscle momentum is extracted. We discriminate an abnormal case and a normal case by using a neural network and fuzzy reasoning to confirm the effectiveness of our approach. Finally, in order to show the effectiveness of the proposed method, we show simulation examples using real images.
315.
Takahiro Ogawa, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Feature Extraction from EEG Patterns in Music Listening, Proc. of International Symposium on Intelligent Signal Processing and communication Systems'2004, S01-4-1-S01-4-5, Seoul, Nov. 2004.
(Summary)
Various illnesses are caused by stress, and stress release is being carried out by musical therapy. A variety of music is used in the musical therapy, and it takes a long time for patient and music therapist to select the music. Generally, time selecting music can be reduced and the musical therapy can be done more easily if effective music for the purpose it is easily found. For this purpose, we measure and extract an EEG (electroencephalogram) difference between music genres as characteristic data in this paper. Our method produces data based on frequency appearance rate, extracts features by principal component analysis, and then analyzes them by using a neural network. Finally in order to show the effectiveness of the proposed method, we carried out computer simulations by using the real data.
316.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Face Recognition using Genetic Algorithm based Template Matching, Proc. of IEEE International Symposium on Communications and Information Technologies'2004, 1252-1257, Sapporo, Oct. 2004.
Yuji Matasumura, Minoru Fukumi, Norio Akamatsu and Fumiaki Takeda : Wrist EMG Pattern Recognition System by Neural Networks and Multiple Principal Component Analysis, Proc. of 8th International Conference on Knowledge-Based Intelligent Information & Engineering Systems'2004, 891-897, Welington, Sep. 2004.
(Summary)
In this paper, we aim for construction of high-speed and high-accurate system using Fast Fourier Transform (FFT) for feature extraction, Simple-PCA (SPCA) for feature compression, and a neural network (NN) for recognition. In particular, we present a novel method based on Multiple PCA to improve recognition accuracy for EMG. From results of computer simulation, it is shown that our approach is effective for improvement in recognition accuracy and speed.
318.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Object Extraction System by Using the Evolutionaly Computations, Proc. of 8th International Conference on Knowledge-Based Intelligent Information & Engineering Systems'2004, 881-890, Welington, Sep. 2004.
(Summary)
In this paper, we propose a new robust thresholds determination method in the various background by using the real coded genetic algorithm (RGA). By using RGA, the most likely plate colors are decided under various light conditions. First, the average brightness Y values of images are calculated. Next, relationship between the Y value and the most likely plate color thresholds (upper and lower bounds) are obtained by GA to estimate threshold equations by using the recursive least squares (RLS) algorithm. Finally, in order to show the effectiveness of the pro-posed method, we show simulation examples by using real images, and result rate of detection is 85.0%.
319.
Takashi Imura, Minoru Fukumi, Norio Akamatsu and Kazuhiro Nakaura : Face Search by Neural Network Based Skin Color Threshold Method, Proc. of 8th International Conference on Knowledge-Based Intelligent Information & Engineering Systems'2004, 840-847, Welington, Sep. 2004.
(Summary)
In recent years, face recognition is becoming important in security or picture search. Many researches on face recognition are premised on acquisition of face position, and face search is therefore important as a preprocessing of face recognition. Although there is a technique of performing face search using skin color information in color images, threshold determination for extracting skin color area is difficult to be done by brightness variance. In this paper, a neural network (NN) is used for this purpose. The thresholds suitable for extracting a skin color area are learned by NN, and face search insensitive to variation in various pictures with a different skin color is performed.
320.
Hiroshi Kawasaki, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Music Compression System Using the GA, Proc. of 8th International Conference on Knowledge-Based Intelligent Information & Engineering Systems'2004, 827-832, Welington, Sep. 2004.
(Summary)
In this paper, we propose a new system identification method by using a genetic algorithm (GA) which has a hybrid structure. The hybrid structure means that a GA has 2 structures. One is the most popular chromosome type GA. That is, chromosomes have binary type genes. The other one is real coded GA. The former is used for determining a function type automatically. The latter is used for determining the coefficient of the function, time delay in the system and combination of the functions automatically. Finally, in order to show the effectiveness of the proposed method, computer simulations were done. Furthermore, in the computer simulations, 2-kinds of systems are identified. One is the hammer stain model. The other is a complex model. From these simulation results, the effectiveness of the proposed method is cleared.
321.
Miyoko Nakano, Fumiko Yasukata and Minoru Fukumi : Age classification from face images focusing on edge information, Proc. of 8th International Conference on Knowledge-Based Intelligent Information & Engineering Systems'2004, Vol.1, 898-904, Wellington, Sep. 2004.
(Summary)
For achieving high-accurate age classification, this study focuses on the edge information that consists of all wrinkles in a face and also a neck. Density histogram in which the value of the edge intensity of the vertical direction and the horizontal direction in an image is greater than a threshold value is computed by using the edge information. They are treated as input data to a neural network. This classification is tried toward the face image database which collected face images of 15-64 years old. In order to show the effectiveness of the proposed method, computer simulations are carried out using these real images.
322.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Genetic Lips Extraction Method with Flexible Search Domain Control, Proc. of 8th International Conference on Knowledge-Based Intelligent Information & Engineering Systems'2004, 799-806, Wellington, Sep. 2004.
(Summary)
In this paper, a lips extraction method by genetic algorithms (GAs), which has simple, high speed and high accuracy properties is described. This method is a visual front end of audio-visual speech recognition systems for mobile devices. Our method is based on template matching with GA. GA can use a significant amount of computation time, because the calculation and test of fitness values for generation × population is necessary. Furthermore GA is basically a global optimization algorithm. So, premature convergence in GA often causes it to fail to reach the global optimum. To extract lips region at high speed with high accuracy, the population must be small and local search must be more efficient. In order to overcome these problems, we propose the method to control search domain flexibly. We compare this method with the previous method by small population. The experimental results show the proposed method is effective.
323.
Yuuki Yazama, Minoru Fukumi, Norio Akamatsu and Yasue Mitsukura : Vowel Recognition Method by Using Features Included in Amplitude for Mobile Device, Proc. of 13th IEEE International Workshop on Robot and Human Interactive Communication'2004, 613-618, Kurasiki, Sep. 2004.
(Summary)
We propose a method of individual adaptive vowel recognition system, which uses features extracted from an amplitude waveform of vowel. We think that a simple vowel recognition system can be constructed by using a shape feature of waveform without converting a frequency. Then, we extract the amplitude features and divide five kinds of vowels into each combination of two kinds of vowels. The suitable amplitude features for vowel identification are selected for each vowel combination. A distribution of the suitable amplitude features is indicated by using discrete Voronoi diagram and sphere of influence of the amplitude feature is decided. A borderline between spheres of influence is identified to a function of vowel identification by using the least squares method. Finally, in order to show the effectiveness of the proposed method, computer simulations of vowel recognition are carried out by using the functions of vowel identification.
kensuke Mitsukura, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Witold Pedrycz : Medical Diagnosis System Using the Intelligent Fuzzy Systems, Proc. of 8th International Conference on Knowledge-Based Intelligent Information & Engineering Systems'2004, 807-826, Welington, Sep. 2004.
(Summary)
We propose the method to develope a system to detect the asynergy in the left ventricle. Processing images are X-ray photograms of the left ventricle by cardiac catheterization. In this paper, we propose the detection system of the asynergy in the left ventricle by using neural networks and the fuzzy inference. Furtheremore, in order to show the effectiveness of the proposed method, we show the simulation example by using the real data.
325.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Feature Extraction for Face Detection and Recognition., Proc. of 13th IEEE International Workshop on Robot and Human Interactive Communication'2004, No.pp. 235-239, Kurashiki, Japan, Sep. 2004.
(Summary)
We propose a facial feature extraction method for face detection and recognition using image segmentation with adaptive thresholds and real coded genetic algorithm guided shape matching. The shapes template is constructed using the average outer edges of the lips and the eyes. Image segmentation is performed using a region growing method, whose seeds are determined using a hybrid method that combines histogram, random and pixel-by-pixel methods. Adaptive thresholds are calculated using color variance. Color spaces used are the YIQ, XYZ and the HIS. Color variance is worked out using square, star and plus kernels.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Recognizing Frontal Faces using Neural Networks., Proc. of 8th International Conference on Knowledge-Based Intelligent Information & Engineering Systems'2004, 1045-1050, Wellington, Sep. 2004.
(Summary)
In this paper, a neural network based face recognition system is presented. One of the main problems encountered when using neural networks for face recognition is luck of enough training data. This is because, in most cases, only one image per subject is available. Therefore, one of our objectives is to solve the problem of luck of enough data to train neural networks. For each image we ''increase'' the data available by several processes for example, mirroring of the image, using color, edges information, etc. The neural network is trained using structural learning to reduce its size. To represent the face color, the YIQ and the XYZ color spaces are used.
327.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Invariant Lips Extraction for Variation of Horizontal Direction, Proc. of the 6th IASTED International Conference on Signal and Information Processing, 58-63, Honolulu, Aug. 2004.
(Summary)
In this paper, as a new visual front end of audio-visual speech recognition systems, a simple method for lips re gion extraction is proposed. This method has robustness for varying shape and three-dimensional geometric changes by using only one image template. The image data is repre sented by x component (redness) of the Yxy colour space. The system environment is that a camera and human head run around separately for mobile devices. In our previous study, we tried to extract lips region, which has varying shape by speech, and the geometric changes of lips on two dimension. Then, we extend this previous method to three dimension by using projective geometry. In our simula tions, we try to extract lips region from a face image with background, varying shape by speech, geometric changes, and face horizontal direction change. By this simulation, we can obtain high extraction accuracy.
328.
Miyoko Nakano, Fumiko Yasukata and Minoru Fukumi : Age and Gender Classification from Face Images Using Neural Networks, Proc. of the 6th IASTED International Conference on Signal and Information Processing, 69-73, Honolulu, Aug. 2004.
(Summary)
In this paper, in order to achieve high-accurate age es-timation, we paid attention to the edges that consist of all wrinkles in a face and also a neck. In particular, this method uses the value of gray scale in an edge image. Therefore, the feature values of gray scale are fed into input units of a neural network for age estimation. In order to show the effectiveness of the proposed method, the proposed age es-timation method was applied to an age estimation system using real images. For the simulation results, the rate of a classification divided by age was approximately 90% as the whole.
329.
Yuji Matsumura, Minoru Fukumi and Norio Akamatsu : Wrist EMG Pattern Recognition System by Neural Networks and Genetic Algorithms, Proc. of the 6th IASTED International Conference on Intelligent Systems and Control'2004, 421-426, Honolulu, Aug. 2004.
(Summary)
In this paper, we aim for construction of a high-speed and high-accurate system using Fast Fourier Transform (FFT) for feature extraction, Simple-PCA (SPCA) for feature compression, and a neural network (NN) for recognition. Furthermore, we reduce the number of units in an input layer of NN using a genetic algorithm (GA) for EMG recognition. From results of computer simulation, it is shown that our approach is effective for improvement in recognition accuracy and speed.
330.
Yuuki Yazama, Minoru Fukumi, Norio Akamatsu and Yasue Mitsukura : Analysis and Recognition of Wrist Motions by Using Multidimensional Directed Information and EMG signal, Proc. of North American Fuzzy Information Processing Society'2004, 867-870, Banff, Jun. 2004.
(Summary)
We propose a recognition system of wrist motions using EMG signals. The objective in this paper is to recognize 7 wrist patterns using EMG signals measured in wrist. Many EMG signal researches have been presented in humanitarian use for handicapped persons. However, almost all the researches were based on measurement from comparatively large muscular region such as arm and shoulder in data measurement location. The EMG signals are measured with 4-channel sensor in wrist and are recorded as 4 channel's time series data. Multi-dimensional directed information (MDI) are applied to the 4 channel's frequency components. The information capacity, which is the amount of information propagation to flow out of a specific signal to other ones, is computed. The information flow is then used as the feature of wrist motions in a preprocessing, followed by a neural network and motion recognition by the EMG signal is carried out. Finally, in order to demonstrate the effectiveness of the proposed method, computer simulations are carried out for the wrist motions recognition.
331.
Yasue Mitsukura, Kayoko Miyata, Kensuke Mitsukura, Minoru Fukumi and Norio Akamatsu : Intelligent Medical Diagnosis System Using the Fuzzy and Neural Networks, Proc. of North American Fuzzy Information Processing Society'2004, 550-554, Banff, Jun. 2004.
(Summary)
Various imaging diagnostic technologies are studied and used in practical. It is necessary to develop the automatic diagnosing processing system for detecting the internal organ. In Japan, cardiac disease is one of the most common cause of death. Therefore, it is necessary to measure cardiac function quantitatively and evaluate the motions of continuous cardiac muscle. Moreover, we propose the developing the system to detect the asynergy in the left ventricle. The processing images are X-ray photograms of the left ventricle by cardiac catheterization. In this paper, we propose the detection system of the asynergy in the left ventricle by using neural networks and the fuzzy inference. Furthermore, in order to show the effectiveness of the proposed method, we show the simulation example by using the real data.
332.
Seiki Yoshimori, Minoru Fukumi, Norio Akamatsu and Yasue Mitsukura : License plate detection system by using threshold function and improved template matching method, Proc. of North American Fuzzy Information Processing Society'2004, 357-362, Banff, Jun. 2004.
(Summary)
License plate recognition is very important in an automobile society. However, it is very difficult to do it, because a background and a car surface color can be similar to that of the license plate. Furthermore, detection of cars moving at a very high-speed is difficult to be done. We propose a new method to extract a car license plate automatically by using a genetic algorithm (GA). By using GA, the most likely plate colors are decided under various light conditions. First, the average brightness Y values of images are calculated. Next, relationship between the Y value and the most likely plate color thresholds (upper and lower bounds) are obtained by GA to estimate threshold equations by using the RLS algorithm. Finally, in order to show the effectiveness of the proposed method, we show simulation examples by using real images.
333.
Hiroshi Nishiyama, Hiroshi Kawasaki, Minoru Fukumi, Norio Akamatsu and Yasue Mitsukura : Rule Extraction from a Trained Neural Network for Image Keywords Extraction, Proc. of North American Fuzzy Information Processing Society'2004, 325-329, Banff, Jun. 2004.
(Summary)
This paper presents a rule extraction method from a trained neural network (NN), which is used for keywords extraction from images. In our approach, first, a bit map image in the RGB color space is transformed into that in the L*a*b* color space. Next, it clusters image pixels using the fuzzy c-means method and domains are extracted through a labeling process. Features, such as area of obtained domains, color information, and coordinates of the center of gravity, are then calculated, which are used as input attributes to NN. NN is then trained using such features. After NN learning, rule extraction is carried out using binarized output values in the hidden layer for each keyword. The rules extracted in this paper are If-then rules, which include logical functions. The methods of generating keywords using NN and the rules are presented and their comparative experiments are performed. Finally the validity of these methods was verified by means of computer simulations.
334.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Morphing Face Images Using Automatically Specified Features, Proc. of Midwest Symposium'2003, 995-1000, Cairo, Dec. 2003.
(Summary)
In this paper, we present a method using which face images can be automatically warped and morphed. Image warping can be defined as a method for deforming a digital image to different shapes. Image morphing combines image warping with a method that controls the color transition in the intermediate images produced. To morph one image to another, new positions and color transition rates for the pixels in each of the images in the sequence must be calculated. Three processes are involved; feature specification, warp generation and transition control.
335.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Accuracy and Speed Improvement in Lips Region Extraction, Proc. of Australian and New Zealand Conference on Intelligent Information Systems'2003, 495-500, Sydney, Dec. 2003.
(Summary)
In this paper, a method for improvement of extraction accuracy and speed of lips region is described. In our previous study, we tried to extract lips region whose shape is varied by speech. To solve these problem, we use template matching and the matching process is performed using a genetic algorithm. In ddition, we think a characteristic shape and colour of the lips are effective for lips region extraction. In our simulations, we obtain more efficient exploration, high accuracy and very short speed processing time. In this paper, these method of improvement and comparisons of results between our conventional method and improved method are described.
336.
Miyoko Nakano, Fumiko Yasukata, Shinobu Matsuo, Minoru Fukumi and Norio Akamatsu : Age Estimation from Face Images by Neural Networks, Proc. of Australian and New Zealand Conference on Intelligent Information Systems'2003, 325-329, Sydney, Dec. 2003.
Hiroshi NIshiyama, Minoru Fukumi and Norio Akamatsu : Rule Extraction from Images Using Neural Networks, Proc. of Australian and New Zealand Conference on Intelligent Information Systems'2003, 293-297, Sydney, Dec. 2003.
(Summary)
In this paper, as a method of extracting keywords, feature data is extracted from an image and a method using a Back-Propagation neural network for learning and classifying them is proposed. In the L*a*b* color space, it clusters image pixels using the c-means method and domains are extracted by Labeling. Features, such as area of the obtained domain, color information, and coordinates of the center of gravity, are then calculated, which are input attributes to a neural network. Moreover, rule generation is carried out by extracting values in the hidden layer of each keyword after the learning of the neural network. Finally the validity of this method was verified by means of computer simulations.
338.
Yuji Matsumura, Minoru Fukumi and Norio Akamatsu : Recognition of Wrist EMG Signal Patterns by Neural Networks, 2003 International Symposium on Intelligent Signal Processing and Communication System(ISPACS 2003), Vol.B5-4, 572-574, Awaji Island,Japan, Dec. 2003.
339.
Yuji Matsumura, Minoru Fukumi and Norio Akamatsu : Recognition of Wrist EMG Signal Patterns by Neyral Networks, Proc. of International Symposium on Intelligent Signal Processing and Communication Systems, 1-6, Awaji-island, Hyogo, Dec. 2003.
Hironori Nagayama, Minoru Fukumi and Norio Akamatsu : A Segmentation Method for Syllable Recognition, Proc. of International Symposium on Intelligent Signal Processing and Communication Systems, 1-5, Awaji-island, Hyogo, Dec. 2003.
Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Feature Extraction of the EEG Using the Factor Analysis and the Neocognitron, Proc. of International Conf. on Control Automation and Systems'2003, 2217-2220, Gyeonju, Korea, Oct. 2003.
(Summary)
It is known that an EEG is characterized by the unique and personal characteristics of an individual. Little research has been done to take into account these personal characteristics when analyzing EEG signals. Often the EEG has frequency components which can describe most of the significant characteristics. These combinations are often unique like individual human beings and yet they have an underlying basic characteristic as well. We think that these combinations are the personal characteristics frequency components of the EEG. In this seminar, the EEG analysis method by using the Genetic Algorithms (GA), Factor Analysis (FA), and the Neural Networks (NN) is proposed. The GA is used for selecting the personal characteristic frequency components. The FA is used for extracting the characteristics data of the EEG. The NN is used for estimating the characteristics data of the EEG. Finally, in order to show the effectiveness of the proposed method, classifying the EEG pattern is carried out via computer simulations. The EEG pattern is evaluated under 4 conditions: listening to Rock music, Schmaltzy Japanese ballad music, Healing music, and Classical music. The results, when personal characteristics frequency components are NOT used, gave over 80 % versus 95 % accuracy when personal characteristics frequency components are used. This result of our experiment shows the effectiveness of the proposed method.
342.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : The Important Frequency Band Selection and Feature Vector Extraction System by an Evolutional Method, Proc. of International Conf. on Control Automation and Systems'2003, 2209-2212, Gyeonju, Korea, Oct. 2003.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Number Plate Detection System by Using the Night Images, Proc. of International Conf. on Control Automation and Systems'2003, 1249-1253, Gyeonju, Korea, Oct. 2003.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Tracking of Moving Object Using Deformable Template, Proc. of Knowledge-Based Intelligent Information & Engineering Systems'2003, 1162-1168, Oxford, Sep. 2003.
(Summary)
A method to track a moving object that has a continuous shape deformations is described in this paper. The deformation of the template and matching process are performed using a genetic algorithm. A new method, which distinguishes between an object function and fitness function, is proposed in this paper. Comparisons between our conventional method and the new method are also simulated. As an object for this simulation, a lips region during speech is used. From results of this simulation, it comes to light that the proposed method is more better than the conventional one in tracking accuracy and speed.
345.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Reconstruction of Facial Skin Color in Color Images, Proc. of Knowledge-Based Intelligent Information & Engineering Systems'2003, 1135-1141, Oxford, Sep. 2003.
(Summary)
In this paper, a method to recover and reconstruct skin color from images over or under-exposed to scene illumination is proposed. The skin color regions most susceptible to over-exposure include areas around the forehead, tip of the nose and the cheeks. This method uses a feed-forward neural network and a distance measure method to learn and construct the skin color face image map. The color distance measure method used is the Cylindrical Metric method (CMM). The relationships between neighboring pixels skin color distance and their relative position in a face are learned using the neural network. This information is then used to reconstruct over or under-exposed skin color regions.
346.
kensuke Mitsukura, Yasue Mitsukura, Minoru Fukumi and Sigeru Omatu : An Age Estimation System Using the Neural Network, Proc. of Knowledge-Based Intelligent Information & Engineering Syst, 1129-1134, Oxford, Sep. 2003.
(Summary)
In this paper, a Genetic Algorithm (GA) is used to select the most likely values of lips and skin colors in a light condition. It is possible to extract objects from the multi-value image only with the color information. In this paper, the objects of extraction are chosen to be the human lips and skin colors. Furthermore, we propose a face decision standard. That is, the decision method of face or not. Furthermore, it is very important to identify an individual. Therefore in this paper, detected faces are distinguished in individual by using the color maps.
347.
Miyoko Nakano, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Fumiko Yasukata : True Smile Recognition Using Neural Networks and Simple PCA, Proc. of Knowledge-Based Intelligent Information & Engineering Systems'2003, 631-637, Oxford, Sep. 2003.
(Summary)
Recently, an eigenface method by using the principal component analysis (PCA) is popular in a filed of facial expressions recognition. In this study, in order to achieve high-speed PCA, the simple principal component analysis (SPCA) is applied to compress the dimensionality of portions that constitute a face. By using Neural Networks (NN), the difference in value of cosθ between true and false (plastic) smiles is clarified and the true smile is discriminated. Finally, in order to show the effectiveness of the proposed face classification method for true or false smile, computer simulations are done with real images.
348.
Yuji Matsumura, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Recognition of EMG Signal Patterns by Neural Networks, Proc. of Knowledge-Based Intelligent Information & Engineering Systems', 623-630, Oxford, Sep. 2003.
(Summary)
This paper tries to recognize EMG signals by using neural networks. The electrodes under the dry state are attached to wrists and then EMG is measured. These EMG signals are classified into seven categories, such as neutral, up and down, right and left, wrist to inside, wrist to outside by using a neural network. The NN learns FFT spectra to classify them. Moreover, we structuralized NN for improvement of the network. It is shown that our approach is effective to classify the EMG signals by means of computer simulations.
349.
Masayuki Shinmoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Neural Network Approach to Color Image Classification, Proc. of Knowledge-Based Intelligent Information & Engineering System, 617-622, Oxford, Sep. 2003.
(Summary)
This paper presents a method for image classification by neural networks which uses characteristic data extracted from images. In order to extract characteristic data, image pixels are divided by a clustering method on YCrCb 3-dimensionl-color space and processed by labeling to select domains. The information extracted from the domains is characteristic data (color information, position information and area information) of the image. Another characteristic data, which is extracted by Wavelet transform, is added to the feature and a comparative experiment is conducted. Finally the validity of this technique is verified by means of computer simulations.
350.
Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Feature Extraction of the EEG Using the Factor Analysis and Neural Networks, Proc. of Knowledge-Based Intelligent Information & Engineering Systems'2003, 609-616, Oxford, Sep. 2003.
(Summary)
In this paper, the EEG analysis method by using the GA, the FA, and the NN is proposed. The GA is used for selecting the personal characteristics frequency compnents. The FA is used for extracting the characteristics data of the EEG. The NN is used for estimating extracted the characteristics data of the EEG. Finally, in order to show the effectiveness of the proposed method, classifying the EEG pattern does computer simulations. The EEG pattern is 4 conditions, which are listening to Rock music, Schmaltzy Japanese ballad music, Healing music, and Classical music. The result, in the case of not using the personal characteristics frequency components, gave over 80 % using the personal characteristics frequency components, gave over 95 % effectiveness of the proposed method.
351.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Feature Extraction Method for Personal Identification System, Proc. of Knowledge-Based Intelligent Information & Engineering Syst, 601-608, Oxford, Sep. 2003.
(Summary)
Recently, personal identification using faces are used because of needless for physical contact. However, when the number of registrant of a system increase, the recognition accuracy of system will get worse certainly. Therefore, in order to improve the recognition accuracy, it is necessary to extract the feature area effectively for getting the high recognition accuracy. In this paper, we analyze and examine about the individual feature in a face using the GA and the SPCA. Thus, by removing the area which is not valuable, we think that recognition accuracy becomes high. Then, in order to show the effectiveness of the proposed method, we show computer simulations by using real image.
352.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Recognition from EMG signals by an Evolutional Method and Non-Negative Matrix Factorization, Proc. of Knowledge-Based Intelligent Information & Engineering Systems'2003, 594-600, Oxford, Sep. 2003.
(Summary)
In this paper, we propose a method of the noise rejection from a signal acquired from many channels using the Electromyograph (EMG) signals. The EMG signals is acquired by the 4th electrodes. EMG signals of 4ch(es) is decomposed into two processions using Non-Negative Matrix Factorization(NMF). And noise rejection is performed by applying the filter obtained by GA to the decomposed matrix . After performing noise rejection, EGM signals is reconstructed and the acquired EMG signal is recognized. The EMG signals based on 7 operations at a wrist are measured. We show the effectiveness of this method by means of computer simulations.
353.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : License Plate Detection using Hereditary Threshold Determined Method, Proc. of Knowledge-Based Intelligent Information & Engineering Systems'2003, Oxford, England, pp.585-593 (2003)., 585-593, Oxford, Sep. 2003.
(Summary)
In this paper, we propose a new thresholds determination method in the various background by using the real-coded genetic algorithm (RGA). By using RGA, the most likely plate colors are decided under various light conditions. First, the average brightness Y values of images are calculated. Next, relationship between the Y value and the most likely plate color thresholds (upper and lower bounds)are obtained by RGA to estimate thresholds function by using the recursive least squares (RLS) algorithm. Finally, in order to show the effectiveness of the proposed method, we show simulation examples by using real images.
354.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Sigeru Omatu : The Image Recognition System by Using the FA and SNN, Proc. of Knowledge-Based Intelligent Information & Engineering Systems'2003, 578-584, Oxford, Sep. 2003.
(Summary)
In this paper, keywords in the image are analized by using the factor analysis and the sandglass-type neural network (SNN) for image searching. As images preprocessing, objective images are segmented by the maximin-distance algorithm. Small regions are integrated into a near region. Thus, objective images are segmented into some region. After this images preprocessing, keywords in images are analyzed by using factor analysis and a sandglass-type neural network (SNN) for image searching in this paper. Images data are corresponded to 2-dimensional data by using these two methods. 2-dimensional data are plotted on a graph. Images are recognized by using this graph.
355.
Hideki Matsuda, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : 3-Dimensional Object Recognition by Evolutional RBF Network, Proc. of Knowledge-Based Intelligent Information & Engineering Systems'2003, 556-562, Oxford, Sep. 2003.
(Summary)
This paper tries to recognize 3-dimensional objects by using an evolutional RBF network. Our proposed RBF network has the structure of preparing four RBFs for each hidden layer unit, selecting based on the Euclid distance between an input image and RBF. This structure can be invariant to 2-dimensional rotation by 90 degree. The other rotational invariance can be achieved by the RBF network. In hidden layer units, the number of RBFs, form, and arrangement are determined using real-coded GA. Computer simulations show object recognition can be done using such a method.
356.
Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : The Recognition Method of the EEG Feature Pattern Using the Factor Analysis, Proc. of 34th Annual Conference of the International Simulation and Gaming Association, 987-996, Chiba, Aug. 2003.
Kensuke Mitsukura, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Sigeru Omatu : Face Extraction and Identification System Using Double Structure Neural Networks, Proc. of 34th Annual Conference of the International Simulation and Gaming Association, 1011-1020, Chiba, Aug. 2003.
Satoshi Tadokoro, Taketsugu Nagao, Minoru Fukumi and Norio Akamatsu : Research on Drift Ice Detection from SAR Images Using Bio-Information Processing, Proc. of the 5th IASTED International Conference on Signal and Information Processing, 167-172, Honolulu, Aug. 2003.
(Summary)
In this paper, drift ice is detected in order to prevent marine accident in winter and improve reliability of sea data. This paper aims at automatic detection of drift ice using NN (Neural Networks) effective in nonlinear processing. Furthermore, GA (Genetic Algorithms) is used for reduction of the network size of NN and acquisition of the data that function effective in drift ice detection.
359.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Improvement of Lips Region Extraction Method for Lipreading, Proc. of the 5th IASTED International Conference on Signal and Information Processing, 127-132, Honolulu, Aug. 2003.
(Summary)
In this paper, a lips extraction method at the mo ment of speech is described. Our method uses template matching and the matching process is performed using a genetic al gorithm. Only one template of closed mouth region is used in consideration for using on mobile devices. Parameters resolved by the genetic algorithm show fine location and geometric change. As an image data, a modified x compo nent (redness) of the Yxy colour space. A new method, which distinguishes between an objective function and fitness function, is proposed in this paper. We simulate comparison between our conventional method and the new method. Experimental results indicate that the new method is more better than the conventional one in extrac tion accuracy and speed.
360.
Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Rajiv Khosla : The EEG Detection System Using the Factor Analysis and Neural Networks, Proc. of 7th World Multi-Conference on Systemics, Cybernetics and Informatics, Vol.IV, 330-333, Orlando, Jul. 2003.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Khosler Rajiv : License Area Extraction System Using the Real-Coded GA, Proc. of 7th World Multi-Conference on Systemics, Cybernetics and Informatics, Vol.IV, 269-274, Orlando, Jul. 2003.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Feature Vector Extraction System from EMG Signals Using Gentic Algorithm, Proc. of 7th World Multi-Conference on Systemics, Cybernetics and Informatics, Vol.IV, 82-87, Orlando, Jul. 2003.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Feature Extraction Method for Personal Identification System by Using Real-Coded Genetic Algorithm, Proc. of 7th World Multi-Conference on Systemics, Cybernetics and Informatics, 66-70, Orlando, Jul. 2003.
(Summary)
本国際会議論文は，実数値型遺伝的アルゴリズムにより個人認証を高精度に行うための特徴抽出方法を提案している．特に顔領域の重要度を実数値型遺伝的アルゴリズムにより決定し，Simple PCAにより特徴を生成し，さらにニューラルネットワークにより学習認識する方法を提案している．100名の被験者(ただし，学習用の被験者は50名)に対する計算機シミュレーションにより，本手法の有効性を定量的に検証している．本国際会議において，Best Presentation Award を受賞した．
364.
Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Feature Extraction of the EEG during Listening to the Music using the Factor Analysis and Neural Networks, Proc. of International Joint Conference on Neural Network, 2263-2267, Portland, Jul. 2003.
(Summary)
Recently in the world, the research of the electroencephalogram (EEG) interface is done, because it has the possibility to realize an interface that can be operated without special knowledge and technology by using the EEG as a means of the interface. As one of the EEG interface, as for a goal for the final of this research, the EEG control system by any music is constructed. However, the EEG control by music is very difficult because it does not know the music and the causal relation of the EEG clearly. Therefore, the EEG analysis and music analysis is absolutely imperative in this system. In this paper, the EEG analysis method by using the FA and the NN is proposed. The FA is used for extracting the characteristics data of the EEG. The NN is used for estimating extracted the characteristics data of the EEG. Moreover teacher signal data of the NN uses the data of the characteristics data of the music. The characteristics data of music is extracted by using the Bark scale analysis. Finally, in order to show the effectiveness of the proposed method, classifying the EEG pattern is done computer simulations. The EEG pattern is 4 conditions, which are listening to Rock music, Schmaltzy Japanese ballad music, Healing music, and Classical music.
365.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Recognition system for EMG signals by using non-negative matrix factorization, Proc. of International Joint Conference on Neural Network, 2130-2133, Portland, Jul. 2003.
(Summary)
IN this paper, the feature vector of a few dimensions for the electromyograph (EMG) recognition systems is extracted. We aim at the construction of the comprehensive operation equipment to which the operation used frequently was summarized. Important frequency bands of EMG signals are selected by using a genetic algorithm. The EMG signals are a kind of the living organism signal. The EMG signals based on 7 operations at a wrist are measured and recognized. We perform a recognition experiment of EMG signals by neural network using the selected frequency band. We show the effectiveness of this method by means of computer simulations.
366.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Face detection and emotional extraction system using double structure neural networks, Proc. of International Joint Conference on Neural Network, 1253-1257, Portland, Jul. 2003.
(Summary)
In this paper, we propose a new method to examine whether or not human faces are included in color images by using a lip detection neural network (LDNN) and a skin distinction neural network (SDNN). First, the lip is detected by LDNN in the proposed method. Next, SDNN is utilized to distinguish the skin color from the others. The proposed method can obtain relatively high recognition accuracy, since it has the double recognition structure of LDNN and SDNN. Finally, in order to demonstrate the effectiveness of the proposed scheme, computer simulations were performed.The validity was verified by testing images containing several faces.
367.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A design of the object detection system using the RGA, Proc. of International Joint Conference on Neural Network, 1227-1231, Portland, Jul. 2003.
(Summary)
License plate recognition is very important is an automobile society. However, it is very difficult to do it, because the background and body color of cars are sometimes similar to that of the license plate. Furthermore, the detection of cars moving at a very high speed is difficult. In this paper, we propose a new robust thresholds-determination method in various backgrounds by using the real coded genetic algorithm (RGA). By using RGA, the most likely plate colors are decided under various light conditions. First, the average brightness Y values of images are calculated. Next, relationship between the Y value and the most likely plate color thresholds (upper and lower bounds) are obtained by GA to estimate threshold equations by using recursive least squares (RLS) algorithm. Finally, in order to show the effectiveness of the proposed method, we show simulation examples by using real images. The resulting rate of detection is 85.0%.
368.
Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Rajiv Khosla : An EEG Feature Detection System Using the Neural Networks Based on Genetic Algorithms, Proc. of IEEE International Symposium on Computational Intelliegnce in Robotics and Automation, 1196-1200, Kobe, Jul. 2003.
(Summary)
In this paper, the EEG analysis method by using the GA, the FA and the NN is proposed. The GA is used for selecting the personal characteristics frequency components. The FA is used for extracting the characteristic data of the EEG. The NN is used for estimating extracted the characteristics data of the EEG. Finally, in order to show the effectiveness of the proposed method, EEG pattern was classified using computer simulations. The EEG pattern has 4 conditions, which are listening to rock music, Schmaltzy Japanese ballad music, healing music and classical music. The result, in the case of not using the personal characteristics frequency components, gave over 95% accuracy. This result of our experiment shows the effectiveness of the proposed method.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Neural Networks and Genetic Algorithms for Learning the Scene Illumination in Color Images, Proc. of IEEE International Symposium on Computational Intelliegnce in Robotics and Automation, 1085-1089, Kobe, Jul. 2003.
(Summary)
This paper proposes the use of neural networks to estimate the scene or camera white balancing illuminant in color images. A real coded genetic algorithm is also used to shirk the size of the neural network input data (thus, a smaller network that will train and run faster), and to identify the areas of a scene that contain the most information about, and therefore best represent the scene illuminant (this offers better data quality thereby improving the accuracy). In addition structural learning with knowledge was incorporated with the back propagation method to further reduce the size of the neural network. The final appearance of the colors in an image depends on many factors. The three main factors are the scene illumination, the illuminant used to white balance the image acquisition device and the properties of the object. While both the camera white balancing illuminant and the properties of the object might be known a prior, it is not always possible to control the scene illumination. In this paper, facial skin color was used as the object interest. Given a visual scene, the proposed method first searches for the presence of at least one face in the scene. If one is found, the genetic algorithm is used to extract the inputs to the neural network, that in turn outputs the estimated scene illuminant. Presently, the proposed method is used to estimated four illuminants in the range 2300K to 6500K.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Feature Extraction Method in Face Image for Personal Identification, Proc. of IEEE International Symposium on Computational Intelliegnce in Robotics and Automation, 1081-1084, Kobe, Jul. 2003.
(Summary)
Recently in the world, many researches of personal identification method using biometrics are widely done. Especially, personal identification using the face is used because of needless for physical contact. In this paper, we propose the new method to identify the personal using the SPCA and the NN. The SPCA has been proposed in order for the PCS to achieve high-speed processing. Moreover, the personal identification system needs to have not only ability to recognize registrant correctly but also ability to reject un-registrant certainly. Therefore, we perform simulation using un-registrant. Furthermore, we analyzed and examined about the individual feature in a face by using the GA and the SPCA. Then, in order to show the effectiveness of the proposed method, we show computer simulations by suing the real image. From these results, we show the effectiveness.
Hideaki Sato, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Thresholds Decision Method for Fast Object Detection Systems, Proc. of IEEE International Symposium on Computational Intelliegnce in Robotics and Automation, 1075-1080, Kobe, Jul. 2003.
(Summary)
In this paper, we propose a new determination method of color thresholds for extracting objects. In order to obtain the optimal color threshold, it is necessary to determine RGB values all together because one color consists of a combination of RGB values. The two optimal thresholds, an upper and a lower, are determined for object extraction by applying genetic algorithm. Moreover, in the proposed method, individuals are evaluated using two fitness functions for obtaining the optimal color thresholds, and initial population is generated with color histograms for fast convergence. Finally, computer simulations are performed, and the result images are shown.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi and Sigeru Omatu : Scene Image Analysis by using the Sandglass-type Neural Network with a Factor Analysis, Proc. of IEEE International Symposium on Computational Intelliegnce in Robotics and Automation, 994-997, Kobe, Jul. 2003.
(Summary)
In this paper, keywords in the image are analysed by using the factor analysis and the sandglass-type neural network (SNN) for image searching. As images preprocessing, objective images are segmented by the maximin-distance algorithm. Small regions are integrated into a near region. Thus, objective images are segmented into some region. After this images preprocessing, keywords in images are analyzed by using factor analysis and a sandglass-type neural network (SNN) for image searching in this paper. Images data are compressed to a 2-dimensional space by using these two methods. This 2-dimensional data space is presented by a graph. Thus, keywords are analyzed in detail.
Takuya Akashi, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Genetic Lips Extraction Method for Varying Shape, Proc. of IEEE International Symposium on Computational Intelliegnce in Robotics and Automation, 982-987, Kobe, Jul. 2003.
(Summary)
In this paper, a lips extraction method that can extract lips region from varying lips shape at the moment of speech by using only one template image is described. The method that is proposed in this paper, has invariance for an open and closed mouth, showing or not showing any teeth, and has high speed and high extraction accuracy in consideration for characteristics of the lips by using the genetic method. This method uses the template matching based on the genetic algorithm. Furthermore, the colour of lips and characteristics of the lips shape variances at the moment of speech in this system are utilized. The effectiveness of this method is demonstrated with only one template for each person being tested and a search object, that is, the varying lip shapes at the moment of speech of vowels by means of computer simulations. These computer simulations indicate that this method can extract the varying lips shape at the moment of speech by using only one template. Moreover, in the extraction processing of every vowel, a high speed and high extraction accuracy can be obtained.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : License Plate Detection System in Rainy Days, Proc. of IEEE International Symposium on Computational Intelliegnce in Robotics and Automation, 972-976, Kobe, Jul. 2003.
(Summary)
License plate recognition is very important in an automobile society. However, it is very difficult to do it, because a background and a body color of cars are similar to that of the license plate. Furthermore, the detection of cars in a moving at a very high-speed is difficult to be done. In this paper, we propose a new robust thresholds determination method in the various background by using the real coded genetic algorithm (RGA). By using RGA, the most likely plate colors are decided under various light conditions. First, the average brightness Y values of images are calculated. Next, relationship between the Y value and the most likely plate color thresholds (upper and lower bounds) are obtained by GA to estimate threshold equations by using the recursive least squares algorithm. Finally, in order to show the effectiveness of the proposed method, we show simulation examples by using real images.
Yasue Mitsukura, Kensuke Mitsukura, Minoru Fukumi, Norio Akamatsu and Sigeru Omatu : Robust Face Detection for Direction Changing Using Evolutionary Algorithms, Proc. of IEEE International Symposium on Computational Intelliegnce in Robotics and Automation, 870-873, Tokyo, Jul. 2003.
(Summary)
Threshold selection in multi-value images is performed based on their color information. When the threshold in an image is fixed it lacks versatility for the others. Because the color information varies under the influence of light conditions. In this paper, a Genetic Algorithm (GA) is used to select the most likely values of lips and skin colors in a light condition. It is possible to extract objects from the multi-value image only with the color information. In this paper, the objects of extraction are chosen to be the human lips and skin colors.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Feature Analysis for the EMG Signals Based on the Class Distance, Proc. of IEEE International Symposium on Computational Intelliegnce in Robotics and Automation, 860-863, Kobe, Jul. 2003.
(Summary)
In this paper, a feature vector is extracted from an electromyography (EMG) signal at a wrist, and the EMG signals based on 7 motions are recognized. In order to perform good pattern recognition, it is desirable that the distance in feature vector between classes is far, and that the variance in a class is small. In consideration of these, important frequency bands of EMG signals are selected by using a genetic algorithm. We use the selected frequency band to perform the recognition experiment of EMG signal by a neural network. Finally, the effectiveness of this method is demonstrated by means of computer simulations.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Identifying Scene Illumination using Genetic Algorithms and Neural Networks, CD-ROM Proc. of 2002 International Conference on Neural Information Processing, 1-5, Singapore, Nov. 2002.
(Summary)
The scene illumination that is present when an image is captured plays a major role in the appearance of the colors of the objects in that image. This is because color is not a physical property of objects and therefore, it is not constant under most conditions, for example, changing scene illumination. We target skin color as the object of interest in an image. We used neural networks to learn and identify which of four different types of scene illuminations were present in a given scene. A real coded genetic algorithm was also used to shrink the size of the neural network input data, and to identify the areas of a scene that contain the most information about, and best represent the scene illuminant Once a scene illuminant can be identified, it can go a long way in helping to correct and normalize skin color in systems' that are sensitive to skin color changes, for example face detection and recognition.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of Face Detection System using the GA and the Simple PCA, CD-ROM Proc. of 2002 International Conference on Neural Information Processing, Vol.4, 2069-2073, Singapore, Nov. 2002.
(Summary)
Recently in the world, many researches of individual recognition method using biometrics are widely done. Especially, individual recognition using faces are used because of needless for physical contact. When we specify with an individual, the information from a face is the most reliable. In this paper, we propose the new method of individual recognition using Simple PCA and NN. Individual recognition system needs to have not only ability to recognize the registrant correctly but also ability to reject the other person certainly. therefore, our objective is to reject certainly the other person. Furthermore, we analyzed and examined about the individual feature in a face using GA. Moreover, in order to show the effectiveness of the proposed method, we show computer simulations by using real image. From these results, we show the effectiveness.
Yuji Matsumura, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Recognition of EMG Signal Patterns by Neural Networks, CD-ROM Proc. of 2002 International Conference on Neural Information Processing, 1-5, Singapore, Nov. 2002.
(Summary)
This paper tries to recognize EMG signals by using neural networks, These EMG signals are classified into seven categories, such as neutral, up and down. right and left, wrist to inside and wrist to outside. The neural network learns FFT spectra to classify them. It is shown that our approach is effective to classify the EMG signals by means of computer simulations.
Masayuki Shinmoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Neural Network Approach to Color Image Classification, CD-ROM Proc. of 2002 International Conference on Neural Information Processing, Vol.2, 675-679, Singapore, Nov. 2002.
(Summary)
The paper presents a method for image classification by neural networks which uses characteristic data extracted from images. In order to extract characteristic data, image pixels are divided by a clustering method on HSI 3-dimensional-color space and processed by labeling to select domains. The information extracted from the domains is characteristic data (color information, position information and area information) of the image. Another characteristic data, which is extracted by Hough transform, is added to the feature and a comparative experiment is conducted. Finally the validity of this technique is verified by means of computer simulation.
Miyoko Nakano, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : True Smile Recognition System using Neural Networks, CD-ROM Proc. of 2002 International Conference on Neural Information Processing, Vol.2, 650-654, Singapore, Nov. 2002.
(Summary)
Recently, research about man-machine interfaces has increased. Therefore application to facial expressions is expected from the development of the man-machine interface. An eigen-face method is popular in these research fields by using the principal component analysis (PCA). But in PCA, it is not easy to compute eigenvectors with a large matrix when considering the cost of calculation to adapt for time-varying processing. In order for PCA to become high-speed, the simple principal component analysis (SPCA) is applied to compress the dimensionality of portions that constitute a face. A value of cos θ is calculated using the eigenvector and the gray-scale image vector of each picture pattern. By using neural networks (NN), the value of cos θ between true and false (plastic) smiles is clarified and the true smile is discriminated. Finally, in order to show the effectiveness of the proposed face classification method for true or false smile, computer simulations are done.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Face Detection and Emotional Extraction System using Double Structure Neural Networks, CD-ROM Proc. of 2002 International Conference on Neural Information Processing, 1-5, Singapore, Nov. 2002.
(Summary)
We propose a new method to examine whether or not human faces are included in color images by using a lip detection neural network (LDNN) and a skin distinction neural network (SDNN). In conventional methods, if there exists the same color as the skin color in scenes, the domain which is accepted as not only the skin color but any other color can be searched. However, first, the lips are detected by LDNN in the proposed method. Next, SDNN is utilized to distinguish skin color from the other colors. The proposed method can obtain relatively high recognition accuracy, since it has the double recognition structure of LDNN and SDNN. Finally, in order to demonstrate the effectiveness of the proposed scheme, computer simulations were performed. First, 100 lip color, 100 skin color and 100 background pictures, which are changed into 10×10 pixels, are prepared for training. The validity was verified by testing images containing several faces.
Taketugu Nagao, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Drift ICE Recognition using Remote Sensing Data by Neural Networks, CD-ROM Proc. of 2002 International Conference on Neural Information Processing, 1-5, Singapore, Nov. 2002.
(Summary)
In recent years, observation of a wide variety in the Earth's surface can be done by improvement of remote sensing technology. The purpose of the paper is to recognize a drift ice as thick ice, thin ice, and sea using synthetic aperture radar (SAR) images. The recognition of the drift ice is achieved by using neural networks (NN). The neural network applies two methods, a BP trained neural network and a self-organizing map. Training data are image features extracted from SAR images. There are three methods for extracting the features: Fourier transform, high-order autocorrelation function (HACF), and image features based on a run length method. We carry out a comparative experiment, and demonstrate their effectiveness by means of computer simulation.
Hideaki Sato, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A proposal of emotional detection system from speech data, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, 1362-1366, Crema, Sep. 2002.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Key words extraction for image search of WWW, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, 1357-1361, Crema, Sep. 2002.
Satoru Tadokoro, Minoru Fukumi and Norio Akamatsu : Drift ice detection in remote sensing images using neural networks and genetic algorithms, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, 353-357, Crema, Sep. 2002.
Yoshihiko Fukuta, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : High-speed face search by threshold value determination method using neural networks, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, 348-352, Crema, Sep. 2002.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Estimating Scene and Camera calibration Illuminants in color images, Proc. of SIP, 266-270, Kuai, Hawaii, Aug. 2002.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A system identification method using a hybrid-type genetic algorithm, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, 364-368, Crema, Jul. 2002.
(Summary)
In this paper, we propose a new system identification method by using a genetic algorithm (GA) which has a hybrid structure. The hybrid structure means that a GA has 2 structures. Finally, in order to show the effectiveness of the proposed method, computer simulations were done. Furthermore, in the computer simulations, 2-kinds of systems are identified. One is the hammer stain model. The other is a complex model. From these simulation results, the effectiveness of the proposed method is cleared.
390.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Lips extraction with template matching by genetic algorithms, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, 343-347, Crema, Jul. 2002.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Skin Color Compensation Under Varying Scene Illumination, Proc. of IASTED International Conference on Artificial Intelligence and Soft Computing, 500-505, Banff, Jul. 2002.
(Summary)
We propose to normalize skin color under varying illumination using a graphical method under any type of scene illumination. Under a given illumination, skin color occupies a well-defined region in certain color systems like YIQ, YCrCb etc and therefore can be easily detected using threshold techniques. When a scene illuminant changes, the skin color changes to a different color, but the skin color can still be detected if the threshold were adjusted. This information can be mapped and used to compensate for skin color changes under any illuminant.
392.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Proposal of Face Decision Standard Using Dual-structured Neural networks, Proc. of Artificial Intelligence and Soft Computing, 397-402, Banff, Jul. 2002.
(Summary)
In this paper, we propose a new method to examine whether or not human faces are included in color images by using a lip detection neural network (LDNN)and a skin distinction neural network (SDNN). First, the lip are detected by LDNN in the proposed method. Next, SDNN is utilized to distinguish the skin color from the others. The proposed method can obtain relatively high recognition accuracy, since it have the double recognition structure of LDNN and SDNN. Finally, in order to demonstrate the effectiveness of the proposed scheme, computer simulations were performed.
393.
Hideaki Sato, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Emotional Space Map Generation Method by Using Neural Networks, Proc. of Artificial Intelligence and Soft Computing, 436-439, Banff, Jul. 2002.
(Summary)
3
394.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Feature Data Selection for Rainfall Forecast by Using Real-Coded GA, Proc. of IASTED International Conference on Artificial Intelligence and Soft Computing, 260-263, Banff, Jul. 2002.
(Summary)
In this paper, rainfall is forecasted by using a Neural Network (NN) and a Genetic Algorithm (GA). GA selects data needed to predict the rainfall. NN learns and forecasts it using attributes selected by GA. The real-coded GA is used to decide data priority degree, and data really needed for the rainfall forecast is selected based on the priority. Finally, in order to show the effectiveness of the proposed fainfall forecast system, computer simulations are performned for real weather data.
395.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Proposal of Impression Extraction System Using a Neural Network and a Genetic Algorithm, Proc. of 6th World Multiconference on Systemics, Cybernetics and Informatics, 380-383, Orlando, Jul. 2002.
Atsuko Ichihara, Yasue Mitsukura, Minoru Fukumi and Motokatsu Yasutomo : An Automatic Liver Extraction Method Using Moving Points and Neural Networks, Proc. of 6th World Multiconference on Systemics, Cybernetics and Informatics, 368-373, Orlando, Jul. 2002.
(Summary)
Automatic extraction of tumors is needed in a medical field. The extraction of a liver field and liver tumors are difficult to be done because the CT images include various internal organs and there is little difference in CT level between a liver field and tumors. In this paper, the extraction method of the liver field and liver tumors is proposed to analyze the information of CT images. A liver field is extracted using the method of moving points, and then the tumors are extracted using neural entworks. It is shown the method has an ability to extract liver tumors.
397.
Yuji Matsumura, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu, Yoshihiro Yamamoto and Kazuhiro Nakaura : EMG Signal Recognition by Using Neural Networks, Proc. of International Symposium on Advanced Control of Industrial Processes, Vol.MA2C-10, 251-256, Kumamoto, Jun. 2002.
(Summary)
This paper tries to recognize EMG signals by using neural networks. The electrodes under the dry state are attached to wrists and then EMG is measured. These EMG signals are classified into seven categories, such as neutral, up and down, right and left, wrist to inside, wrist to outside by using a neural network. The NN learns FFT spectra to classify them. Moreover, we structuralized NN for improvement of the network. It is shown that our approach is effective to classify the wrist EMG signals by means of computer simulations
398.
Yasue Mitsukura, Minoru Fukumi and Toru Yamamoto : A Design of System Identification Method by Using a Hybrid Genetic Algorithm, Proc. of International Symposium on Advanced Control of Industrial Processes, 145-148, Kumamoto, Jun. 2002.
Yasue Mitsukura, Seiji Ito, Minoru Fukumi and Norio Akamatsu : Genetic Fog Occurrence Forecasting System Using a LVQ Network, Proc. 20th IASTED Int. Multi-Conf. on Applied Informatics (AI2002), 200-203, Innsbruck, Feb. 2002.
Yasue Mitsukura, Minoru Fukumi and Toru Yamamoto : Design and Experimental Evaluation of Self-Tuning Intelligent PID Controller using Genetic Algorithm, Proc. 20th IASTED Int. Multi-Conf. on Applied Informatics (AI2002), 476-482, Innsbruck, Feb. 2002.
(Summary)
PID control scheme have been widely used in most of process control systems for the control of chemical processes. However, the selection of suitable or optimal tuning parameters for such control laws remains a challenging problem. The PID constants have a great influence on stability and control performance. In this paper, a self-tuning PID control scheme is proposed, which is able to deal with a time varying system. The proposed scheme is derived based on the relationship betweeb PIS control and generalized minimum variance control (GMVC) laws. Furthermore, a suitable set of some user-specified parameters included in the GMVC criterion is sought by using a genetoc algorithm recursively. FInally, the newly proposed scheme is evaluated on a temperature control system.
401.
Yasue Mitsukura, Hideaki Sato, Minoru Fukumi and Norio Akamatsu : Fast Face Detection System for Time-Varying Images Using the GA-Threshold Method, Proc. 20th IASTED Int. Multi-Conf. on Applied Informatics (AI2002), 89-92, Innsbruck, Feb. 2002.
(Summary)
Threshold selection in multi-value images is performed based on their color information. When the threshold in an image is xed it lacks versatility for the others. Because the color information varies under the in uence of light conditions. In this paper, a Genetic Algorithm (GA) is used to select the most likely values of lips and skin colors in a light condition. It is possible to extract objects from the multi-value image only with the color information. In this paper, the objects of extraction are chosen to be the human lips and skin colors.
402.
Miyoko Nakano, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Recognition of Facial Expressions Based on Simple PCA, Proc. 20th IASTED Int. Multi-Conf. on Applied Informatics (AI2002), 63-67, Innsbruck, Feb. 2002.
(Summary)
In this paper, the simple principal component analysis (SPCA) is applied to dimensionality compression of por tions that constitute a face, which is a data-oriented fast method. An angle (cos θ) is calculated using the eigenvec tor and the gray scale image vector of each picture pattern. By using the value of cos θ, similarity between true and false (plastic) smiles is clari ed and the true smile is dis criminated. Finally, in order to demonstrate the effective ness of the proposed face smile or false classifying method, computer simulations are done.
403.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Rainfall Forecast Using a Neural Network with a Real Coded Genetical Preprocessing, Proc. 20th IASTED Int. Multi-Conf. on Applied Informatics (AI2002), 210-213, Innsbruck, Feb. 2002.
(Summary)
In this paper, rainfall is forecasted by a Neural Network (NN) and a Genetic Algorithm (GA). GA selects data needed to predict the rainfall. NN learns and forecasts it using attributes selected by GA. The real-coded GA is used to decide data priority degree, and data really needed for the rainfall forecast is selected based on the priority. Finally, in order to show the effectiveness of the proposed rainfall forecast system, computer simulations are performed for real weather data.
404.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Rotational Invariant Human Faces Detection in Color Images, Proc. 20th IASTED Int. Multi-Conf. on Applied Informatics (AI2002), 78-83, Innsbruck, Feb. 2002.
(Summary)
This paper presents a skin color filter based detection method that can detect faces in one of the three possible orientations. Since skin color is rotational invariant, we propose the use of a Skin-color Edge Angular Measurement (SEAM) method. This method finds the edges of the skin color and then uses the least square (LS) method to draw straight-line segments along the edges. After pairing the line segments, the angle each of them makes with the horizontal plane is then calculated. Using this calculation, it is possible to determine the direction of orientation of the faces, re-rotate them to a frontal pose and then perform face detection. This filter is fast, with a detection accuracy of 75.8% on our test set.
405.
Hideaki Sato, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of Emotional Classification System and Emotional Space Map, Proc. 20th IASTED Int. Multi-Conf. on Applied Informatics (AI2002), 294-297, Innsbruck, Feb. 2002.
(Summary)
In this paper, we propose a new emotional speech classi cation method and new space map generation method by analyzing feature parameters obtained from the emotional speech and using neural networks. In many cases, emotions are found by the change of pitches. Input values of neural networks (NNs) are then emotional pitch patterns, which are time-varying. It is shown that NNs can achieve emotional speech classi cation and generation of space map by learning each emotional pitch pattern by means of computer simulations. Finally, many tests for emotional speech classi cation are done by using the obtained emotional space map.
406.
Eisaku Ohta, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Searching of Liver Ranges in CT Images by Using Genetic Algorithms and Neural Networks, Proc. 20th IASTED Int. Multi-Conf. on Applied Informatics (AI2002), 84-88, Innsbruck, Feb. 2002.
(Summary)
Recently, internal human organ disorders that medical image analysis can be used to detect is being actively re searched. The research, however, has concentrated on the extraction of pulmonary tumors. There is, therefore, little research being done on the extraction of liver tumors. This is because there is no difference between concentrated values of a healthy part and one with a tumor in liver CT images. In this paper, the extraction method of such liver tumors is proposed. Furthermore, in order to demonstrate the effectiveness of the proposed scheme, we show simulation examples, using real CT image data.
407.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Detection of Human Faces in Visual Scenes, Proc. of Seventh Australian and New Zealand Intelligent Information Systems Conference, 165-170, Perth, Western Australia, Nov. 2001.
(Summary)
This paper presents a neural network based high-speed system to detect faces in any visual scene. This paper proposes a high-speed and accurate method to search for human faces using a "Knowledge" pruned small sized neural network, and skin colour detection using threshold method with confirmation by a skin colour detection neural network (TSCD). This project is made up of two parts: the face detecting system (FDS) and the TSCD. The FDS that is used to detect faces is made up of a knowledge pruned face locator, a down sampler, and a merger. The TSCD does a high-speed reduction of the face search area to skin regions (and then to face candidates). The TSCD assumes, correctly, that a human face (without wearing any mask or painting) in a visual scene can only be found in a skin colour region. However, all skin regions containing faces must be found otherwise the overall system accuracy goes down.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Fast Face Detection System Using the GA-based Threshold Method, Proc. of Seventh Australian and New Zealand Intelligent Information Systems Conference, 183-187, Perth, Western Australia, Nov. 2001.
(Summary)
Threshold selection in multi-value images is performed based on their color information. When the threshold in an image is fixed it lacks versatility for the others. Because the color information varies under the influence of light conditions. In this paper, a genetic algorithm (GA) is used to select the most likely values of lips and skin colors in a light condition. It is possible to extract objects from the multi-value image only with the color information. In this paper, the objects of extraction are chosen to be the human lips and skin colors.
Hideaki Sato, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Emotional Speech Classification with Prosodic Prameters by Using Neural Networks, Proc. of Seventh Australian and New Zealand Intelligent Information Systems Conference, 395-398, Perth, Nov. 2001.
(Summary)
Interestingly, in order to achieve a new Human Interface such that digital computers can deal with the KASEI information, the study of the KANSEI information processing recently has been approached. In this paper, we propose a new classification method of emotional speech by analyzing feature parameters obtained from the emotional speech and by learning them using neural networks, which is regarded as a KANSEI information processing. In the present research, KANSEI information is usually human emotion. The emotion is classified broadly into four patterns such as neutral, anger, sad and joy. The pitch as one of feature parameters governs voice modulation, and can be sensitive to change of emotion. The pitch is extracted from each emotional speech by the cepstrum method. Input values of neural networks (NNs) are then emotional pitch patterns, which are time-varying. It is shown that NNs can achieve classification of emotion by learning each emotional pitch pattern by means of computer simulations.
Eisaku Ohta, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : An Extraction Method of Liver Tumors by Using Genetic Algorithms and Neural Networks, Proc. of Seventh Australian and New Zealand Intelligent Information Systems Conference, Perth, Western Australia, Nov. 2001.
(Summary)
Recently, internal human organ disorders that medical image analysis can be used to detect is being actively researched. The research have however, concentrated on the extraction of pulmonary tumors. There is therefore, little research being done on the extraction of liver tumors. This is because there is no difference between concentrated values of a healthy part and one with a tumor in liver CT images. In this paper, the extraction method of such liver tumors is proposed. Furthermore, in order to demonstrate the effectiveness of the proposed scheme, we show a simulation example, using real CT image data.
Minoru Fukumi, Yasue Mitsukura and Norio Akamatsu : A Knowledge Processing Method in Neural Networks, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, Vol.1, 1493-1498, Osaka, Oct. 2001.
Hideaki Sato, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Classification of Emotional Speech by Using Neural Networks, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, Vol.1, 1009-1014, Osaka, Oct. 2001.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of the Fast Face Detection System by the GA-Based Threshold, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, Vol.1, 1015-1020, Osaka, Oct. 2001.
Minoru Fukumi, Yasue Mitsukura and Norio Akamatsu : Knowledge Incorporation and Rule Extraction in Neural Networks, Proc. of International Conference on Artificial Neural Networks, 1248-1253, Wien, Aug. 2001.
(Summary)
In this paper a new knowledge incorporation and rule extraction method in neural networks is presented. The rule form of an if-then type can be inserted into a neural network (NN) as knowledge of a problem. NN is then trained by using a set of training samples. In this case the structure learning algorithm with forgetting is used to generate a small-sized NN system. After the NN training, rules are extracted from it. The results of computer simulations show that this approach can generate obvious network architectures and as a result simple rules compared with conventional rule extraction methods.
415.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Detection Method of Face Regions in Color Images by Using Evolutionally Computation, Proc. of International Joint Conference on Neural Networks, 2253-2257, Washington, D.C., Jul. 2001.
(Summary)
This paper presents a new scheme to design a face detection system by using a genetic algorithm (GA). In particular, the object of this paper is to design a fast face detection system. 5 human emotions are considered; neutrality, happiness, sadness, anger and surprise. This paper pays attention to lip forms in human characteristics. Furthermore, in order to decrease the detection time, GA is used for the purpose of focusing on the lips in images. Finally, in order to demonstrate the effectiveness of the proposed scheme, we show simulation examples.
Atsuko Ichihara, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Automatic Extraction of Tumors by Using Neural Computation, Proc. of International Joint Conference on Neural Networks, 2981-2984, Washington, D.C., Jul. 2001.
(Summary)
Automatic extraction of tumors is needed in a medical field. A 3D image is composed of the sequence of 2D CT images. The extraction of a liver field is difficult, because the CT images include various internal organs. In this paper, an improved active net is proposed to analyze the information of CT images. It is shown this method has the ability as a pre-processing system for extraction of liver tumors.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Rainfall Forecast Using a Neural Network with a Genetic Preprocessing, Proc. of Knowledge- Based Intelligent Information Engineering Systems & Allied Technologies, Vol.1, 812-817, Osaka, Jan. 2001.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Rainfall Forecast Using a Neural Network and a Genetic Algorithm, Utilization of Soft Computing Techniques for Intelligent Prediction, 812-817, 2001.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Improved Searching Speed for Human Face Detection in Visual Scenes using Neural Networks, Proc. of International Conference on Neural Information Processing, No.FBP-30, 1-6, Taejon, Nov. 2000.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of Face Detection System in Color Images by Using Evolutionary Computation, Proc. of International Conference on Neural Information Processing, 1126-1130, Korea, Nov. 2000.
(Summary)
We propose a new method to examine whether or not human faces are included in color images by using a lip detection neural network (LDNN) and a skin distinction neural network (SDNN). In conventional methods, if there are the same color as the skin color in scenes, the domain which is accepted as not only the skin color but any other color can be searched. However, first, the lip are detected by LDNN in the proposed method. Next, SDNN is utilized to distinguish the skin color from the others. The proposed method can obtain relatively high recognition accuracy, since it have the double recognition structure of LDNN and SDNN. Furthermore, in order to reduce the amount of calculation and to discriminate a skin color and the other colors, GA can search to minimize the number of necessary data. Then a storage capacity and the amount of operations can be cut down by using GA and LDNN and SDNN are trained by using the reduced data. Finally, in order to demonstrate the effectiveness of the proposed scheme, computer simulations were performed. First, 100 lip color, 100 skin color and 100 background pictures, which are changed into 10×10 pixels, are prepared for training. The validity was verified by testing images containing several faces.
421.
Motomichi Inoue, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Neural Net based Image Retrieval by Using Color and Location Information, Proc. of the 2000 IEEE International Conference on Systems, Man and Cybernetics, 2575-2579, Nashville, Oct. 2000.
(Summary)
A neural net-based image retrieval method is presented, in which color and location features are extracted from images. This method can retrieve similar images to a selected one from a large data set of color images. In particular, the location features in the color distribution of an image are important in the image retrieval. This image retrieval method extracts color features and their location information included in an image. A neural network tries to find images with similar features from a data set. First, images are translated into gray-scale ones and then are divided into eight regions based on gray-scale values. The color and location features are extracted from these regions after integration of regions. The RGB and HSV color values in each region, area, and the X- and Y-values in the orthogonal coordinates are learned by a multi-layered neural network. After learning, the neural network evaluates the similarity between a selected image and the other ones in the data set. The similar images found by the neural network are the retrieval results.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of Face Detection System by Using the Lip Detection Neural Network and Skin Distinction Neural network, Proc. of the 2000 IEEE International Conference on Systems, Man and Cybernetics, 2789-2793, Nashville, Tennessee, Oct. 2000.
(Summary)
The authors propose a method to examine whether or not human faces are included in color images by using a lip detection neural network (LDNN) and a skin distinction neural network (SDNN). In conventional methods, if there are colors the same as skin color in scenes, the domain which is accepted as not only the skin color but any other color can be searched. However, first, the lips are detected by LDNN in the proposed method. Next, SDNN is utilized to distinguish skin color from other colors. The proposed method can obtain relatively high recognition accuracy, since it has the double recognition structure of LDNN and SDNN. Finally, in, order to demonstrate the effectiveness of the proposed scheme, computer simulations were performed. First, 100 lip color, 100 skin color and 100 background pictures, which were changed into 10×10 pixels, were prepared for training. The validity was verified by testing images containing several faces.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of Genetic Fog Occurrence Forecasting System by Using LVQ Network, Proc. of the 2000 IEEE International Conference on Systems, Man and Cybernetics, 3678-3681, Nashville, Tennessee, Oct. 2000.
(Summary)
A transportation development in recent years is quite remarkable. However, poor visibility often cause an accident. Therefore, it is very important to forecast a fog occurrence. In this paper, we propose a scheme to forecast a fog occurrence by using the Learning Vector Quantization (LVQ) and a Genetic Algorithm (GA). This scheme forecasts the fog occurrence by the weather data which are provided from the Japan Meteorological Agency. First, the provided data formation are shown. Next, the prediction scheme is described in detail. In this method, input attributes for a LVQ network are selected by real-coded GA to improve forecast accuracy. Furthermore, a partial selection processing in the real-coded GA improves its convergence properties. Finally, in order to show the effectiveness of the proposed prediction scheme, computer simulations are performed.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of Face Detection System Using Evolutionary Computation, Vol.2, 398-402, Kuala Lumpur, Sep. 2000.
(Summary)
We propose a new method to examine whether or not human faces are included in color images by using a lip detection neural network (LDNN) and a skin distinction neural network (SDNN). In conventional methods, if there are the same color as the skin color in scenes, the domain which is accepted as not only the skin color but any other color can be searched. However, first, the lip are detected by LDNN in the proposed method. Next, SDNN is utilized to distinguish the skin color from the others. The proposed method can obtain relatively high recognition accuracy, since it have the double recognition structure of LDNN and SDNN. Furthermore, in order to reduce the amount of calculation and to discriminate a skin color and the other colors, GA can search to minimize the number of necessary data. Then a storage capacity and the amount of operations can be cut down by using GA and LDNN and SDNN are trained by using the reduced data. Finally, in order to demonstrate the effectiveness of the proposed scheme, computer simulations were performed. First, 100 lip color, 100 skin color and 100 background pictures, which are changed into 10×10 pixels, are prepared for training. The validity was verified by testing images containing several faces.
Minoru Fukumi, Yasue Mitsukura and Norio Akamatsu : A New Genetic Approach to Universal Rule Generation from Trained Neural Networks, Proc. of The IEEE Region 10 International Conference on Electrical and Electronic Technology (TENCON 2000), Vol.1, 1-6, Kuala Lumpur, Sep. 2000.
(Summary)
A new rule generation method from neural networks is presented. A neural network (NN) is formed using a genetic algorithm (GA) with virus infection and deterministic mutation to represent regularities in training data. This method utilizes a modular structure in GA. Each module learns a different neural network architecture, such as sigmoid and a higher order neural networks. Those chromosome information is communicated to the other modules by the virus infection. The higher order units are connected to an output unit or hidden units. By using these architectures, rules can be extracted. The results of computer simulations show that this approach can generate obvious network architectures and as a result simple rules.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of Face Detection System Using an Evolutionary Computation, Proc. of The IEEE Region 10 International Conference on Electrical and Electronic Technology (TENCON 2000), Vol.2, No.2, 553-559, Kuala Lumpur, Sep. 2000.
(Summary)
We propose a new method to examine whether or not human faces are included in color images by using a lip detection neural network (LDNN) and a skin distinction neural network (SDNN). First, the lip are detected by LDNN in the proposed method. Next, SDNN is utilized to distinguish the skin color from the others. The proposed method can obtain relatively high recognition accuracy, since it have the double recognition structure of LDNN and SDNN. Furthermore, in order to reduce the amount of calculation and to discriminate a skin color and the other colors, GA can search to minimize the number of necessary data. Finally, in order to demonstrate the effectiveness of the proposed scheme, computer simulations were performed. The validity was verified by testing images containing several faces.
Minoru Fukumi, Yasue Mitsukura and Norio Akamatsu : A New Rule Generation Method from Neural Networks Formed Using a Genetic Algorithm with Virus Infection, Proc. of International Joint Conference on Neural Networks, No.42-03, 1-6, Como, Jul. 2000.
(Summary)
In this paper a new rule generation method from neural networks is presented. A neural network is formed using a genetic algorithm (GA) with virus infection and deterministic mutation to represent regularities in training data. This method utilizes a modular structure in GA. Each module learns a different neural network architecture, such as sigmoid and a high order neural networks. Those information is communicated to the other modules by the virus infection. The results of computer simulations show that this approach can generate obvious network structures and lead to simple rules.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Design and Evaluation of Neural Networks for Coin Recognition by Using GA and SA, Proc. of International Joint Conference on Neural Networks, Vol.4, 1-6, Como, Jul. 2000.
(Summary)
In this paper, we propose a method to design a neural network (NN) by using a genetic algorithm (GA) and simulated annealing (SA). And also, in order to demonstrate the effectiveness of the proposed scheme, we apply the proposed scheme to a coin recognition example. In general, as a problem becomes complex and large-scale, the number of operations increases and hardware implementation to real systems (coin recognition machines) using NNs becomes difficult. Therefore, we propose the method which makes a small-sized NN system to achieve a cost reduction and to simplify hardware implementation to the real machines. The coin images used in this paper were taken by a cheap scanner. Then they are not perfect, but a part of the coin image could be used in computer simulations. Input signals, which are Fourier spectra, are learned by a three-layered NN. The inputs to NN are selected by using GA with SA to make a small-sized NN. Simulation results show that the proposed scheme is effective to find a small number of input signals for coin recognition.
Yasue Mitsukura, Toru Yamamoto, Masahiro Kaneda, Minoru Fukumi, Norio Akamatsu and Shah L. Sirish : Design of a Self-Tuning PID Controller Using Evolutionary Computation and Its Experimental Evaluation, Proc. of IFAC Symposium PID'00, Vol.3, No.2, 642-647, Terassa Spain, Apr. 2000.
(Summary)
PID control schemes have been widely used in most process control systems represented by chemical processes for a long time. However it is a very important problem how to find the suitable control parameters, i.e., PID parameters, because these parameters give a great influence on the stability and the control performance. In this paper, a self-tuning PID control scheme is proposed, which is able to deal with a time-varying system. The proposed scheme is derived based on the relationship betweeen PID control and generalized minimum variance control (GMVC) laws. Furthermore, a suitable set of some user-specified parameters included in the GMVC criterion is sought by using a genetic algorithm recursively. Finally, the newly proposed scheme is evaluated on a temperature control system.
430.
Minoru Fukumi and Norio Akamatsu : Rule Generation from a Rotation-Invariant Neural Pattern Recognition System, Proc. of International Conference of Neural Information Processing, Vol.II, 706-711, Perth, Nov. 1999.
(Summary)
A method of extracting rules from a rotation-invariant neural pattern recognition system formed using a genetic algorithm (GA) is presented. In particular, deterministic mutation (DM) is utilized to improve its convergence properties. It is performed on the basis of the result of neural network structure learning. DM can evolve chromosomes of individuals to increase their fitness functions in a deterministic manner. In this paper, coin data are used as inputs. The coins used are a Japanese 500-yen coin and a South Korean 500-won coin, which are very similar. GA is utilized to reduce the number of connection weights in the neural network. The network weights surviving after training represent rules to perform pattern classification for the coin data. The rules are then extracted from the network. Furthermore, the network has a procedure to substitute signum units for hidden sigmoid ones in examining its recognition accuracy. It enables us to easily extract rules. Simulation results show that this approach can generate a simple network structure and, as a result, simple rules for coin data classification.
Hideaki Ishii, Nobuyoshi Takeuchi, Minoru Fukumi and Norio Akamatsu : Face detection based on skin color information in visual scenes by neural networks, Proc. of 1999 IEEE International Conference on Systems, Man and Cybernetics, Vol.V, 350-355, Tokyo, Nov. 1999.
(Summary)
A method to examine whether or not human faces are included in the images and to detect their position by using the technique of skin color region extraction is presented. In this technique, the skin color which is a main feature of faces is detected, a binary image composed of skin color parts and background one is constructed from an original image using a neural network which learns color information, and then the skin color parts of some sizes are regarded as face candidates. Thus search regions are limited within the skin color parts. Therefore, an improvement in the detection speed is achieved. These face candidates are examined using a neural network which learns the features of faces, and estimates whether or not the original image includes the faces. From results of computer simulations, a search rate of 83.3 % accuracy was achieved from 15 sheets, each having from 1 to 3 faces. The sizes and positions of faces were chosen as randomly as possible. There was no search of other objects other than faces.
Yasuomi Inooka, Minoru Fukumi and Norio Akamatsu : Learning and analysis of facial expression images using a five-layered hourglass-type neural network, Proc. of 1999 IEEE International Conference on Systems, Man and Cybernetics, Vol.V, 373-376, Tokyo, Nov. 1999.
(Summary)
In this study, a method to perform feature extraction and image creation support of five facial human expressions in gray scale images are presented in which the five-layered hourglass-type neural network is used. Input values to the neural network are five facial expression images composed of 100×100 pixels and are the same as the teacher-signals in the output layer. The teacher-signals are learned using the five-layered hourglass-type neural network to achieve a compression function and restoration of the images. The compressed information and causality of each facial expression are dealt with in the third layer (the emotion-layer). Furthermore, it can be shown that an image creation is performed by giving adequate values to the emotion-layer.
433.
Minoru Fukumi and Norio Akamatsu : An Evolutionary Approach to Rule Generation from Neural Networks, Proc. of The 8th IEEE International Conference on Fuzzy Systems (FUZZ-IEEE'99), Vol.3, 1388-1393, Seoul, Aug. 1999.
(Summary)
A method of extracting rules from neural networks formed using an evolutionary algorithm is presented. The evolutionary algorithm used here is a random optimization method (ROM). In particular, deterministic mutation (DM) is introduced in ROM. It is performed on the basis of the result of neural network learning. The DM procedure can evolve a candidate of a solution to increase a ROM fitness function in a deterministic manner. In the paper iris data are used as inputs. ROM are utilized to reduce the number of connection weights in the neural network. The network weights survived after the ROM training represent rules to perform pattern classification for the iris data. The rules are then extracted from the networks in which hidden units use signum and sigmoid functions to produce binary outputs. It enables us to extract simple logical functions from the network. Simulation results show this approach can generate a simple network structure and as a result simple rules.
Minoru Fukumi and Norio Akamatsu : A New Rule Extraction method from Neural Networks, Proc. of International Joint Conference on Neural Networks, Vol.6, 4134-4138, Washington, D.C., Jul. 1999.
(Summary)
This paper presents a method of extracting rules from multilayered neural networks (NN) formed using a random optimization (search) method (ROM). The objective of this study is to extract rules from NN, achieving 100% recognition accuracy in a pattern recognition system. NNs to be extracted rules are formed using ROM. A hybrid algorithm of NN and ROM performs a formation of a small-sized NN system, which is suitable for a rule extraction. In this paper iris data is used as inputs. ROM is utilized to reduce the number of connection weights in NN. The network weights survived after the ROM training represent regularities to perform pattern classification. The rules are then extracted from the networks in which hidden units use signum and sigmoid functions to produce binary outputs. It enables us to extract simple logical functions from the network. By means of computer simulation, the effectiveness of this approach is examined.
Yasuomi Inooka, Minoru Fukumi and Norio Akamatsu : Numeral Recognition Using a Multi Neural Network Model, Proc. of 1998 International Conference of Neural Information Processing, Vol.2, 1094-1097, Kokura, Sep. 1998.
Kazuhiro Satomi, Minoru Fukumi and Norio Akamatsu : Rule Extraction from Small-Sized Neural Networks Formed Using a Genetic Algorithm, Proc. of 6th International Conference on Soft Computing, Vol.2, 644-647, Iizuka, Sep. 1998.
Minoru Fukumi and Norio Akamatsu : Rule Extraction from Neural Networks Trained Using Evolutionary Algorithms with Deterministic Mutation, Proc. of International Joint Conference on Neural Networks, 686-689, Alaska, May 1998.
(Summary)
A method of extracting rules from neural networks trained using evolutionary algorithms (EAs) is presented. The EAs used are a genetic algorithm (GA) with deterministic mutation (DM) and a random optimization method (ROM) with DM. The DM is performed on the basis of the result of neural network learning. It can evolve chromosomes of individuals to increase their fitness functions in a deterministic manner. The EAs are utilized to reduce the number of neural network connections. The network connections surviving after training represent rules to perform pattern classification. The rules are then extracted from the network in which hidden units use signum functions to produce binary outputs. Simulation results show this method can generate a simple network structure and as a result simple rules for the iris data classification.
Minoru Fukumi and Norio Akamatsu : Designing a Neural Network Using Evolutionary Algorithms with Deterministic Mutation, Proc. of IFAC/IEEE Int. Symposium on AI in RTC, 120-125, Kuala Lumpur, Sep. 1997.
Minoru Fukumi and Norio Akamatsu : A Method to design a Neural Pattern Recognition System by Using a Genetic Algorithm with Partial Fitness and a Deterministic Mutation, Proc. of 1996 IEEE International Conference on Systems, Man, and Cybernetics, 1989-1993, Beijing, Oct. 1996.
(Summary)
This paper presents a method using a genetic algorithm (GA) with a partial fitness (PF) and a deterministic mutation (DM) to design a neural pattern recognition system for a rotated coin recognition problem. In the method, chromosomes in the GA are divided into several parts. Their PFs are evaluated for GA operations. Furthermore, this paper introduces the DM based on a neural network learning. A coin recognition system in this paper includes as a preprocessor the Fourier transform, which produces rotation invariant features. Those features are recognized by a multilayered neural network. The GA is utilized to reduce the number of input signals, Fourier spectra, into the neural network. It is shown that the present method is better than conventional GAs on convergence in learning and makes a small-sized neural network.
Minoru Fukumi and Norio Akamatsu : A Genetic Approach to Feature Selection for Pattern Recognition Systems, Proc. of 4th International Conference on Soft Computing, 907-910, Iizuka, Sep. 1996.
Minoru Fukumi and Norio Akamatsu : A Neural Network for Recognizing Rotated Patterns and Estimating Their Rotation Angle, Proc. of 1996 International Conference on Neural Information Processing, 365-370, Hong Kong, Sep. 1996.
(Summary)
This paper considers a rotation invariant neural pattern recognition system, which can recognize rotated patterns and estimate their rotation angle. It consists of a preprocessing network to detect edge features of input patterns and a trainable multilayered network. The multilayered network has two output layers, which are a part to recognize an input pattern and a part to estimate its rotation angle. The hidden layer is commonly used by both parts. As a result, it can deal with rotation-sensitive and -insensitive nformation. It is shown that, by means of computer simulations on a coin recognition problem, the system is able to recognize rotated patterns and estimate their rotation angle.
442.
Minoru Fukumi, Sigeru Omatu and Yoshikazu Nishokawa : Designing a Neural Network by a Genetic Algorithm with Partial Fitness, Proc. of Int. Conf. on Neural Networks, 1834-1838, Perth, Dec. 1995.
(Summary)
This paper presents a method of using the genetic algorithm (GA) with partial fitness (PF) to design a neural network for coin recognition. The method divides a chromosome in the GA into several parts, the PFs of which are evaluated for GA operations. Each part independently performs selection and crossover operations in the GA. Such a technique improves performance in learning of the GA. This paper applies the method to a rotated coin recognition problem to examine its effectiveness. The coin recognition system described consists of a preprocessor with Fourier transform and a multilayered network. The method is utilized to reduce the number of input signals, Fourier spectra, of the multilayered network. It is shown that the method is better than the conventional GA on convergence in learning and makes a smaller size network.
Minoru Fukumi, Sigeru Omatu and Yoshikazu Nishikawa : Rotation Invariant Neural Pattern Recognition System which Can Estimate a Rotation Angle, Proc. of Int. Conf. on Neural Networks, 4390-4395, Orlando, Jul. 1994.
(Summary)
This paper presents a rotation invariant neural pattern recognition system, which can recognize a rotated pattern and estimate a rotation angle. The system is very effective for a rotated coin recognition problem, but is poor compared with human performance. It is well known that human sometimes recognizes a rotated pattern by means of the mental rotation. Such a fact, however, has never been considered and used in neural pattern recognition systems, especially in rotation invariant systems. Therefore, we examine the principle of mental rotation and apply it to a rotation invariant pattern recognition system. The system with such a principle could recognize a rotated pattern and estimate a rotation angle. It is shown that the system is effective to recognize a rotated pattern from results of computer simulation for a coin recognition problem.
Minoru Fukumi and Sigeru Omatu : Designing A Neural Network for Coin Recognition by A Genetic Algorithm, Proc. of Int. Joint Conf. Neural Networks, Vol.3, 2109-2112, Nagoya, Nov. 1993.
(Summary)
This paper presents a method to design a neural network for coin recognition by a genetic algorithm (GA). The GA specifies an architecture of neural network, but does not train the network. The back-propagation (BP) method trains the network. After training it by the BP, the GA varies the architecture of the network to fit the environment, which is to achieve a 100% recognition accuracy and to make the network small in size. The network reduced by the GA is further decreased by using the BP with forgetting of weight. The object of this paper is to design a smaller neural network for hardware implementation of coin recognition system. Results by computer simulation show the effectiveness of the method to variably rotated coin recognition problem.
Minoru Fukumi, Sigeru Omatu, Fumiaki Takeda and Tishihisa Kosaka : Rotation Invariant Neural Pattern Recognition System with Application to Coin Recognition, Proc. of Int. Joint Conf. on Neural Netwroks, Vol.2, 1027-1032, Singapore, Nov. 1991.
(Summary)
The authors propose a pattern recognition system which is insensitive to the rotation of the input pattern by various degrees. The system consists of a fixed invariance network with many slabs and a trainable multilayered network. To illustrate the effectiveness of the system, the authors apply it to rotation-invariant coin recognition of 500 yen and 500 won coins. The results of computer simulation show that a neural network approach will be useful in rotation-invariant pattern recogniti.
Sigeru Omatu, Minoru Fukumi and Masaru Teranishi : Neural Network Model for Alphabetical Letter Recognition, Proc. of International Neura Network Conference, 19-22, Paris, Jul. 1990.
Sigeru Omatu, Norihiko Ono and Minoru Fukumi : Development of Decision Support System for the Environmental and Energy Problem by the Simulation Game, Proc. of IFAC/IFORS/IAEE International Symposium on Energy System Management and Economics, 403-408, Tokyo, Nov. 1989.
Minoru Fukumi and Sigeru Omatu : A New Back-Propagation Algorithm with Coupled Neuron, Proc. of Int'l Session of the 28th SICE Annual Conference, 1327-1330, Matsuyama, Jul. 1989.
Sigeru Omatu, Naofumi Hosokawa and Minoru Fukumi : A New Approach for Pattern Recognition by Neural Networks with Scramblers, Proc. of International Joint Conference on Neural Networks, 183-188, Washington, D.C., Jun. 1989.
(Summary)
An approach to pattern recognition that is based on a concept involving an invariance net and a trainable classifier is proposed. The invariance net plays an important role in producing a set of outputs that are invariant to translation, rotation, scale change, perspective change, etc., of the retinal input pattern. The trainable classifier is used to classify the scrambled data into the original patterns by using a backpropagation algorithm. The sigmoid functions are adopted as nonlinear elements in the neural networks by B. Widrow et al.'s MRII based on signum functions. Some numerical results are illustrated to show the effectiveness of the present algorithm for pattern recognition.
Minoru Fukumi and Sigeru Omatu : A New Back-Propagation Algorithm with Coupled Neuron, Proc. of International Joint Conference on Neural Networks, 611, Washington, D.C., Jun. 1989.
(Summary)
A novel algorithm is developed for training multilayer fully connected feedforward networks of coupled neurons with both signoid and signum functions. Such networks can be trained by the familiar backpropagation algorithm since the coupled neuron (CONE) proposed uses the differentiable sigmoid function for its trainability. The CONE takes advantages of the key ideas of conventional methods. By applying CNR to a simple network, it is shown that the convergence of the output error is much faster than that of the BP method when the variable learning rate is used. Finally, simulation results illustrate the effective learning algorithm.
Toshiya Morisue and Minoru Fukumi : 3-D Eddy Current Calculation Using the Magnetic Vector Potential, Proc. of IEEE Workshop on Electromagnetic Field Computation,IEEE Schenectady Section, 11-15, Schenectady, Oct. 1986.
Naohiro Okubo, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : 筋電に基づく手首の疲労の検知と動作識別, Proceedings of 2023 Annual Conference of Electronics, Information and Systems Society, IEE of Japan, 1567-1570, Sep. 2023.
2.
Tohma Nakagawa, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Effective Feature Selection in Behavior Identification by EMG, Proceedings of 2023 Annual Conference of Electronics, Information and Systems Society, IEE of Japan, 1000-1002, Sep. 2023.
3.
Ryota Miyake, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Estimate Presence or Absence of Learning Understanding Based on Analysis of EEG and HRV, Proceedings of 2023 Annual Conference of Electronics, Information and Systems Society, IEE of Japan, 983-987, Sep. 2023.
4.
Daiki Fujiwara, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Modeling of Safety Confirnmation Behavior at Intersections and Identifying Distracted State using Time Series Data Learning Methods, Proceedings of 2023 Annual Conference of Electronics, Information and Systems Society, IEE of Japan, 1017-1022, Sep. 2023.
5.
Kazuki Yoshinaga, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Identification of Learning Progress Based on Skeletal Information for Seated Learners, Proceedings of 2023 Annual Conference of Electronics, Information and Systems Society, IEE of Japan, 1011-1016, Sep. 2023.
6.
Ryo Otsuki, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Recognition of Kanji Characters by Aerial Input Using LeapMotion and Personal Authentication, Proceedings of 2023 Annual Conference of Electronics, Information and Systems Society, IEE of Japan, 1003-1006, Sep. 2023.
7.
Katsumasa Nitta, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Estimation of Mood Change using Smartwatch for Depressive State Detection, 情報処理学会第85回全国大会講演論文集, 4-223-4-224, Mar. 2023.
MANA Tahora, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Japanese Sign Language Identification Using Deep Learning with Leap Motion, 電気学会電子·情報·システム部門大会論文集, 1165-1169, Sep. 2022.
𠮷川 京汰, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Development of a smart glass based grilling evaluation system for biginners, 電気学会電子·情報·システム部門大会論文集, 628-632, Aug. 2022.
(Keyword)
集約画像 / deep learning / 手話認識 / カラーリング
10.
畠中 健斗, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Development of a smart glass based grilling evaluation system for biginners, 電気学会電子·情報·システム部門大会論文集, 617-622, Aug. 2022.
(Keyword)
焼き加減 / smart glass / ヒストグラム / 料理 / 初心者
11.
小柳 功王, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Measure Respiration Rate in a Smartwatch Based Sleep Monitoring System, 電気学会電子·情報·システム部門大会論文集, 658-663, Aug. 2022.
(Keyword)
sleep / smartwatch / respiration / ピーク値 / 移動平均
12.
YUKI Saitoh, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : 手首筋電に基づくタッピング動作の認識, 人工知能学会全国大会, 1-4, Jun. 2022.
(Keyword)
deep learning / machine learning / 手首筋電
13.
Shin-ichi Ito, Miura Takanori, Momoyo Ito and Minoru Fukumi : A Method to Detect a Mood Matching Music Using EEG, 電気学会電子·情報·システム部門大会論文集, 601-604, Sep. 2021.
(Keyword)
electroencephalogram / mood / music / machine learning
14.
Shin-ichi Ito, Hironori Kadowaki, Momoyo Ito and Minoru Fukumi : Personal Authentication with Walking Motion Based on Gathered Images and Neural Networks, 電気学会電子·情報·システム部門大会論文集, 512-515, Sep. 2021.
(Keyword)
personal authentication with walking motion / gathered image / deep learning / convolutional neural networks
15.
Ren Nozaki, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : 脳波を用いた面倒な作業に対する感情の検出, 電気学会電子·情報·システム部門大会論文集, 644-648, Sep. 2021.
CHUNYU GUO, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Personal authentication by walking motion using Kinect, 電気学会電子·情報·システム部門大会論文集, 719-721, Sep. 2019.
(Keyword)
human sensing / machine learning / Kinect
27.
Shan Xian, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Study on Discrimination of Finger Motions Based on EMG Signals, 電気学会電子·情報·システム部門大会論文集, 715-718, Sep. 2019.
(Keyword)
human sensing / machine learning / electromyogram activity
28.
Takeru Wasaki, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Direction Discrimination by Vehicle Lamp, 電気学会電子·情報·システム部門大会論文集, 861-865, Sep. 2019.
(Keyword)
machine learning / Vehicle Lamp
29.
Kazuki Shimamoto, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Nail Image Analysis Method for Accumulation Stress Evaluation, 電気学会電子·情報·システム部門大会論文集, 855-860, Sep. 2019.
(Keyword)
human sensing / Nail image / Accumulation Stress
30.
Takuya Kimura, Momoyo Ito, 佐藤 和人, Shin-ichi Ito and Minoru Fukumi : Estimation of Moving Direction of Objects in Driving Scene, 電気学会電子·情報·システム部門大会論文集, 1374-1377, Sep. 2018.
(Keyword)
移動物体識別 / 運転シーン / machine learning / 単眼カメラ
31.
Tomoyuki Sakabe, Momoyo Ito, 佐藤 和人, Shin-ichi Ito and Minoru Fukumi : Identification of Driver Status based on Difference of Visibility using Sparse Structure Learning, 電気学会電子·情報·システム部門大会論文集, 1368-1373, Sep. 2018.
(Keyword)
ドライバー状態 / スパース構造学習 / 視認性
32.
Teru Ando, Minoru Fukumi, Momoyo Ito and Shin-ichi Ito : Persona Identification in monitoring system using Kinect, 電気学会電子·情報·システム部門大会論文集, 1342-1345, Sep. 2018.
(Keyword)
human sensing / machine learning / Kinect
33.
Yurika Fujii, Minoru Fukumi, Momoyo Ito and Shin-ichi Ito : Detection of Dangerous Objects by Pan-tilt Camera, 電気学会電子·情報·システム部門大会論文集, 1516-1517, Sep. 2018.
(Keyword)
危険物検出 / パンチルトカメラ / machine learning
34.
Shion Morikawa, Minoru Fukumi, Momoyo Ito and Shin-ichi Ito : Personal Authentication Using lip EMG by Dry Electrode, 電気学会電子·情報·システム部門大会論文集, 1571-1572, Sep. 2018.
(Keyword)
electromyogram activity / machine learning / Deep learning
35.
Kamat Rahayu Binti Seri, Rahman Arfaus A Mohamad, Zailan Faldi Bin Zul, Minoru Fukumi and Teruaki Ito : Development of ergonomic monitoring system for safe assembly task in manufacturing, 日本機械学会生産システム部門研究発表講演会2018・講演論文集, Vol.18, No.4, 75-76, Mar. 2018.
36.
Ani Firdaus Mohamad, Minoru Fukumi, Kamat Rahayu Binti Seri, Minhat Mohamad and Teruaki Ito : A construction framework of decision support system for improving driving fatigue, 日本機械学会生産システム部門研究発表講演会2018・講演論文集, Vol.18, No.4, 73-74, Mar. 2018.
37.
Shun Yamamoto, Shin-ichi Ito, Momoyo Ito, Minoru Fukumi and Kamat Rahayu Binti Seri : Recognition of Aerial Numerals by Leap Motion and CNN, National Convention Record I.E.E. Japan, 173, Mar. 2018.
(Keyword)
neural network / Leap Motion / CNN / human sensing
38.
Shion Morikawa, Minoru Fukumi, Momoyo Ito, Shin-ichi Ito and Kamat Rahayu Binti Seri : Personal Authentication Using EMG by Dry Electrodes, National Convention Record I.E.E. Japan, 185, Mar. 2018.
(Keyword)
electromyogram activity / SVM / human sensing
39.
Ryosuke Takabatake, Minoru Fukumi, Momoyo Ito, Shin-ichi Ito and Kamat Rahayu Binti Seri : Dataset making for Japanese Vowels Recognition using Surface electromyogram measured with Bipolar Dry Type Sensors, National Convention Record I.E.E. Japan, 180-181, Mar. 2018.
(Keyword)
electromyogram activity / SVM
40.
Konomi Takeyasu, Momoyo Ito, Kazuhito Sato, Shin-ichi Ito and Minoru Fukumi : Detection of Inattentive State Based on Change of Driving Behavior: Study on Driving Behavior Model using GGM, Conference record of HUMAN COMMUNICATION GROUP SYMPOSIUM 2017, HCG2017-I-1-3, Dec. 2017.
(Summary)
In recent years traffic fatality accidents caused by accidental driving are increasing. If it is possible to detect an inattentive state, not only reduction of traffic fatal accidents but also state monitoring of drivers for takeover in automatic operation is realized. In this research, we model driving behavior measured by a driving simulator based on GGM and sparse structural model. In addition, we discus usefulness of the generated driving behavior model.
41.
Tomoyuki Sakabe, Momoyo Ito, Kazuhito Sato, Shin-ichi Ito and Minoru Fukumi : Analysis of Inattentive State by Driving Information using Sparse Structure Learning: Consideration Based on Difference in Visibility, Conference record of HUMAN COMMUNICATION GROUP SYMPOSIUM 2017, HCG2017-C-3-3, Dec. 2017.
(Summary)
The aimless driving is one of the most common cause of the traffic accidents. If we can detect a change of driver's behavior according to the aimless driving, the detection will achieve prevention of traffic accidents and safety takeover of autonomous driving. In this paper, we analyze a deference between normal condition and distracted condition by using sparse structure learning based on correlation of driver's information. We use eye gaze, face direction, heartbeat, vehicle velocity, acceleration, and tilt of handle. We analyze the change of driving information according to the visibility.
42.
FIRDAUS Mohammad, KAMAT Rahayu Seri, MINHAT Mohamad, Minoru Fukumi and Teruaki Ito : Effect of vibration towards drivning fatigue and development of regression model based on vibration, 日本機械学会・第27回設計工学システム部門講演会2017・講演論文集, Vol.17, No.32, 2506-1-2506-10, Sep. 2017.
(Summary)
This paper present the result of whole-body vibration (WBV) of the Malaysian drivers driving a car through different road conditions; straight, winding, uphill, and downhill) at constant speed (80km/h). Driving fatigue has been defined as a feeling of drowsiness due to extending the driving period, type of road condition, adverse climatologically environment or drivers' individual characteristic are the direct contributing factor to road accidents. The objective of this study is to study the effect of the WBV towards the driving fatigue. Besides, the regression modeling of WBV for drivers fatigue was developed. The model can predict the relationship between the process input parameters and output response. There were ten healthy and experienced drivers served as the subjects of this study. The WBV measurement was taken and evaluated using the tri-axial seat pad accelerometer and 4-channel VI-400PRO Human Vibration Meter (HVM). Design Expert 8.0.6 software was used for the development of regression model. This study is expected to analyze the WBV, and develop the regression model of the WBV by using regression analysis. The result of this study indicates that the subjects recorded the vibration values that show they feel fairly uncomfortable as it in the caution zone. The vibration exposure can cause the changes in body chemistry and metabolism, which can lead to fatigue effects. Besides, the regression model was successfully developed and validated. The modeling validation runs were within the 90% prediction interval of the developed model and the residual errors compared to the predicted values were less than 10%. Through this study, the significant parameters that influenced the WBV were also identified. WBV was influenced by the time exposure, type of road, gender, the interaction between time exposure and type of road, and interaction between time exposure and gender. Thus, the author believes there is a new contribution to the body of knowledge throughout this study.
43.
KAMAT Rahayu Seri, HALWANI Nurul, Minoru Fukumi and Teruaki Ito : Design a manual culf massager for prolonged standing workers by using ergonomic approach, 日本機械学会・第27回設計工学システム部門講演会2017・講演論文集, Vol.17, No.32, 2505-1-2505-10, Sep. 2017.
44.
Tanaka Tomoya, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Evaluate Stress Level by Using Expression Analysis, 電気学会電子情報システム部門大会論文集, 1136-1139, Sep. 2017.
(Keyword)
表情分析 / ストレス評価 / アミラーゼ
45.
Omae Hisaki, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method of Evaluate Understanding of Learning Using Electroencephalogram, 電気学会電子情報システム部門大会論文集, 464-467, Sep. 2017.
(Keyword)
脳波 / 信号処理 / 簡易脳波計 / 学習理解
46.
Higasa Takashi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Character Input System using Gesture Motion in Augmented Reality, 電気学会電子·情報·システム部門大会論文集, 1683-1684, Sep. 2017.
(Keyword)
Augmented Reality / HSV color system / Gesture motion / Character string detection / Optical character reader
47.
Taisei Watanabe, Tadahiro Oyama and Minoru Fukumi : Estimation of Tongue Motion and Silent Speech Based on EMG from Suplahyoid Muscles Using CNN, 電気学会電子情報システム部門大会論文集, No.PS3-3, 1553-1554, Sep. 2017.
(Keyword)
electromyogram activity / CNN
48.
Teru Ando, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Respiration detection in monitoring system using Kinect, 電気学会電子情報システム部門大会論文集, No.PS3-7, 1561-1562, Sep. 2017.
(Keyword)
Kinect / SVM
49.
Ryouhei Shioji, Minoru Fukumi, Momoyo Ito and Shin-ichi Ito : Personal Authentication Based on Wrist EMG Analysis by a Convolutional Neural Network, 電気学会情報システム研究会, Vol.IS-17, 57-61, Aug. 2017.
Ryousuke Takabatake, Minoru Fukumi, Shin-ichi Ito and Momoyo Ito : Japanese Vowel Recognition Using Surface Electromyogram Measured with Bopolar Dry type Sensors, 電気学会情報システム研究会, Vol.IS-17, 23-27, Aug. 2017.
(Keyword)
electromyogram activity / SVM
51.
Momoyo Ito, Kazuhito Sato, Shin-ichi Ito and Minoru Fukumi : Safety Confirmation Behavior Analysis Based on Driver's Posture Change, 日本機械学会 第25回交通・物流部門大会(TRANSLOG2016)講演論文集/ 1306, Nov. 2016.
52.
Taiki Nonoguchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A System to Detect and Track Mosquitoes, 平成28年度電気学会電子・情報・システム部門大会講演論文, 412-415, Sep. 2016.
(Keyword)
Mosquito
53.
Daiki Konishi, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : 空中署名と深層学習によるバイオメトリクス認証, DIA'2016, IS2-A10-1-IS2-A10-7, Mar. 2016.
Ryosuke Naitoh, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : 画像特徴量とサポートベクターマシンを用いた運転シーン分類に関する検討, DIA'2016, ISC-C8-1-ISC-C8-4, Mar. 2016.
(Summary)
交通事故の主要因の一つは漫然運転であり，発生箇所が最も多いのは交差点である．ドライバが漫然運転をしている時，内部状態(注意散漫状態，焦り，眠気など)に変化があると考えられる．また，内部状態は車両前方の情景(運転シーン)の影響を受けると考えられる．つまり，運転行動を解析するためには運転シーンがどのような状況であるかを理解する必要がある．そこで，本研究では運転シーンの理解を目的とし，画像特徴量を用いた運転シーン分類について検討する．具体的には，運転シーン画像をBag of Keypoints(BoK)により表現し，分類器にサポートベクターマシン(SVM)を用いて直線区間と交差点を分類した．その結果，78.2%の精度で分類可能であった．
56.
Hiroshi Aoki, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Pose Estimation of Hand for AR, 平成27年度電気関係学会四国支部連合大会講演論文集, 188, Sep. 2015.
57.
Takahiro Toyokawa, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Relevance Analysis of Driver's Gaze and Gaze Object, 平成27年度電気関係学会四国支部連合大会講演論文集, 239, Sep. 2015.
58.
syu Tamura, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Method to Best Music Selection System by EEG, 平成27年度電気学会電子・情報・システム部門大会講演論文, 379-384, Aug. 2015.
(Keyword)
music analysis / spectrum analysis / Fourier transform
59.
Takuma Ogawa, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Improvement of ST-Patch Features for Detection of Abandaned Object, 平成27年度電気学会電子・情報・システム部門大会講演論文, 1490-1491, Aug. 2015.
(Keyword)
pan-tilt camera / ST-Patch feature / HOG feature / Real AdaBoost
60.
Daiki Hiraoka, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Discrimination of Japanese Janken by Support Vector Machine Based on Electromyogram of Wrist, 平成27年度電気学会電子・情報・システム部門大会講演論文, 1514-1515, Aug. 2015.
Momoyo Ito, Kazuhito Sato and Minoru Fukumi : A Time - series Analysis of Head Posture based on Traffic Events, Proceedings of Society of Automotive Engineers of Japan Spring Conference, 2232-2235, May 2015.
(Summary)
This study aims to estimate drivers state using driving behaviors. In this paper, we set a bicycle near-miss event on driving simulator: sudden appearance of bicycle. We examine safety verification behaviors associated with near-miss events at nonregulated intersections with poor visibility. From an assessment of a drivers head movements associated with the sudden appearance of bicycle encountered while approaching a nonregulated intersection, we attempt to analyze the causal relation of the changes of frequency analysis results of safety verification behaviors before and after near-miss events.
62.
Masashi Iwase, Minoru Fukumi and Koji Kashihara : EMアルゴリズムに基づく静脈解析のためのAndroidアプリケーション開発, 信学技法(KBSE研究会), Vol.KBSE2014-48, 55-60, Jan. 2015.
63.
Keiichi Tsuzuki, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Basic Study for Categorization of Driving Scene using Deep Learning, Proceedings of 2014 Shikoku-Section Joint Convention Record of the Institute of Electrical and Related Engineers, 185, Sep. 2014.
Ryosuke Oka, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Extraction of Safety Confirmation Behavior Based on Drivers Posture, Proceedings of 2014 Shikoku-Section Joint Convention Record of the Institute of Electrical and Related Engineers, 184, Sep. 2014.
菅 祥雄, Minoru Fukumi and Koji Kashihara : Automatic Detection of ST Segment Depression on the ECG, 平成26年度電気関係学会四国支部連合大会 講演論文集, 230, Sep. 2014.
66.
Shin-ichi Ito, Momoyo Ito, Shoichiro Fujisawa and Minoru Fukumi : Method for EEG Pattern Classification Considering Human Character, 平成26年電気学会電子・情報・システム部門大会講演論文集, 637-640, Sep. 2014.
67.
Momoyo Ito, Kazuhito Sato and Minoru Fukumi : Analysis of Time-Series Changes of Head Posture for Estimation of Drivers States, Proceedings of Forum on Information Technology 2014, Part 3, 256-258, Sep. 2014.
Akiko Sugiyama, Momoyo Ito, Kazuhito Sato, Shin-ichi Ito and Minoru Fukumi : Difference Analysis of Safety Verification Behavior Based on Near-Miss Event Using Head Motion, Proceedings of 2014 JSAE Annual Congress (Spring), Vol.11-14, 13-18, May 2014.
Momoyo Ito, Kazuhito Sato, Shin-ichi Ito and Minoru Fukumi : Influence Analysis of Near-Miss Event on Change of Safety Verification Behavior, Proceedings of 2014 JSAE Annual Congress (Spring), Vol.12-14, 7-10, May 2014.
森本 健太, Minoru Fukumi and Koji Kashihara : 運転時の縦断勾配錯視における脳活動の検討, 2013年度計測自動制御学会四国支部学術講演会 講演論文集, 67-68, Nov. 2013.
71.
Yusuke Yamamura, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : 足首の筋電による足首の動作の分類, 日本生体医工学会中国四国支部大会, 11, Oct. 2013.
72.
Zhang Peng, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : LVQとEOG法を用いる眼電マウスの開発, 日本生体医工学会中国四国支部大会, 34, Oct. 2013.
73.
Takanori Suzuki, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Detection of Drivers Distraction based on Sparse Structure Learning, 平成25年度電気関係学会四国支部連合大会講演論文集, 219, Sep. 2013.
Taito Mori, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Extraction of Gaze Targets from Driving Scene using Saliency Map, 平成25年度電気関係学会四国支部連合大会講演論文集, 217, Sep. 2013.
Kazuya Yaegashi, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : An Analysis of Changes between Biological Information and Driving Behavior from Traffic Event, 平成25年度電気関係学会四国支部連合大会講演論文集, 221, Sep. 2013.
Koji Miyai, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Feature Extraction of Car Front View for Categorization of Driver Scene, 平成25年度電気関係学会四国支部連合大会講演論文集, 216, Sep. 2013.
Akiko Sugiyama, Momoyo Ito, Shin-ichi Ito and Minoru Fukumi : Evaluation of Scenarios on Driving Simulator for Analysis of Safety Verification, 平成25年度電気関係学会四国支部連合大会講演論文集, 222, Sep. 2013.
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : BCIのための簡易計測装置を用いた生体信号の解析, 電気学会C部門大会, GS13-3, Sep. 2013.
79.
Takahiro Horiuchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : 遺伝的アルゴリズムを用いたパノラマ画像の生成, 電気学会電子·情報·システム部門大会論文集, OS10-3, Sep. 2013.
80.
Peng Zhang, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Recognition of Continuous Eye Motions Using Learning Vector Quantization and EOG-feature Based Methods, 電気学会電子·情報·システム部門大会論文集, GS12-2, Sep. 2013.
Takako Ikuno, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Abandoned Object Detection by Genetic Algorithm with Local Search, 電気学会電子·情報·システム部門大会論文集, OS4-6, Sep. 2013.
Yohei Takeuchi, Momoyo Ito and Minoru Fukumi : Iterative Discriminant Analysis in Non-linear Space, IEICE Technical Report, Vol.112, No.69, 59-64, May 2012.
88.
Momoyo Ito, 佐藤 和人 and Minoru Fukumi : Evaluation of Self Mapping Characteristics for Quantification of Head Motion, IEICE Technical Report, Vol.112, No.69, 17-20, May 2012.
Takuya Akashi, 若佐 裕治, 田中 幹也, Stephen Githinji Karungaru and Minoru Fukumi : 遺伝的アルゴリズムを用いた口唇の3 次元情報の取得, FANシンポジウム2008, 11-14, Oct. 2008.
115.
Tadahiro Oyama, Stephen Githinji Karungaru, Satoru Tsuge, Yasue Mitsukura and Minoru Fukumi : Simple-FLDAにおける追加型アルゴリズムの提案, 第18回インテリジェント・システム・シンポジウム(FAN 2008), 21-24, Oct. 2008.
116.
Tadahiro Oyama, Satoru Tsuge, Stephen Githinji Karungaru, Yasue Mitsukura and Minoru Fukumi : 判別分析における高速な追加学習アルゴリズムの一提案, 電気学会 電子・情報・システム部門大会 講演プラグラム集, 286-289, Aug. 2008.
117.
Choge Hillary, Stephen Githinji Karungaru, Satoru Tsuge and Minoru Fukumi : A Fourier-Space Palmprint Feature Extraction Method for Personal Identification, 電気学会 電子・情報・システム部門大会 講演プラグラム集, 303-307, Aug. 2008.
118.
滝本 裕則, Yasue Mitsukura and Minoru Fukumi : 顔印象解析における顔部位の重要性解析, 電子情報通信学会総合大会, Mar. 2008.
119.
Hironori Takimoto, 桑野 翼, 深井 寛修, Yasue Mitsukura and Minoru Fukumi : 顔特徴が人の年齢知覚に及ぼす影響の解析, 電子情報通信学会総合大会, Mar. 2008.
120.
Minoru Fukumi, Stephen Githinji Karungaru, 中野 実代子, Satoru Tsuge, Takuya Akashi and Yasue Mitsukura : 超高速学習型統計アルゴリズムによる特徴抽出, シミュレーション&ゲーミング学会秋期大会, Oct. 2007.
中野 実代子 and Minoru Fukumi : 学習型統計アルゴリズムによる顔情報処理, 電気学会電子·情報·システム部門大会論文集, Sep. 2007.
127.
深井 寛修, 滝本 裕則, Yasue Mitsukura and Minoru Fukumi : 低解像度画像による見た目年齢推定の一考察, Proceedings of the Japan Industry Applications Society Conference, Aug. 2007.
128.
横松 恵理子, Shin-ichi Ito, Yasue Mitsukura, 曹 建庭 and Minoru Fukumi : 因子分析を用いた嗜好の取得, Proceedings of the Japan Industry Applications Society Conference, II_407-410, Aug. 2007.
129.
村上 純子, Shin-ichi Ito, Yasue Mitsukura, 曹 建庭 and Minoru Fukumi : 脳波による周波数特徴抽出システム, Proceedings of the Japan Industry Applications Society Conference, II_403-406, Aug. 2007.
130.
Shin-ichi Ito, Yasue Mitsukura, 宮村 浩子 and Minoru Fukumi : 脳波に混入するアーチファクトの検出, Proceedings of the Japan Industry Applications Society Conference, II_95-98, Aug. 2007.
131.
近藤将 之, Stephen Githinji Karungaru, Satoru Tsuge, Minoru Fukumi and Yasue Mitsukura : ニューラルネットワークによる画像へのキーワード付加, Proceedings of the Japan Industry Applications Society Conference, II_83-86, Aug. 2007.
132.
Minoru Fukumi, Stephen Githinji Karungaru, 明石 卓也, 中野 実代子 and Yasue Mitsukura : 新統計的学習アルゴリズムによる超高速特徴生成, 高速信号処理応用技術学会2007年研究会, 27, Jul. 2007.
133.
Satoru Tsuge, 清田 啓二, Masami Shishibori, Kenji Kita, Fuji Ren, Minoru Fukumi and Shingo Kuroiwa : Analysis of Variation on Intra-Speaker's Speech Recognition Performances, 日本音響学会 春季研究発表会, 165-166, Mar. 2007.
134.
Satoru Tsuge, 清田 啓二, Masami Shishibori, Kenji Kita, Fuji Ren, Minoru Fukumi and Shingo Kuroiwa : Analysis of Variation on Intra-Speaker's Speech Recognition Performances, 日本音響学会 春季研究発表会, 165-166, Mar. 2007.
Satoru Tsuge, 小澤 光広, Masami Shishibori, Minoru Fukumi, Fuji Ren, Kenji Kita and Shingo Kuroiwa : A Japanese Specific Speakers' Corpus Long Period, 日本音響学会2006年秋季研究発表会, 277-278, Sep. 2006.
138.
竹内 洋平, Minoru Fukumi, Norio Akamatsu, Stephen Githinji Karungaru and 小澤 誠一 : Character Recognition for Malaysian License Plate, システム制御·情報学会講演会, 671-672, May 2006.
139.
松村 悠司, Minoru Fukumi and Yasue Mitsukura : Construction of Hybrid Recognition System by MDA and NN, システム制御·情報学会講演会, 421-422, May 2006.
140.
中道 功, Stephen Githinji Karungaru, Minoru Fukumi, Yasue Mitsukura and 安友 基勝 : Extraction of the Liver Tumor in CT Image by GA, システム制御·情報学会講演会, 667-668, May 2006.
141.
小川 宜洋, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Feature Extraction in Listening to Music Using Analysis of the EEG, システム制御·情報学会講演会, 405-406, May 2006.
142.
Takuya Akashi, 若佐 裕治, 田中 幹也, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Eye Detection by Genetic Algorithms, システム制御·情報学会講演会, 599-600, May 2006.
Miyoko Nakano and Minoru Fukumi : Age and Gender Classification by NN Using Facial Edge Information, Journal of Japanese Academy of Facial Studies Kaogaku, Vol.5, No.1, 172, Oct. 2005.
Miyoko Nakano, Fumiko Yasukata and Minoru Fukumi : A Method of Age Classification Focusing on Edge Information, Proceedings of the 48th Annual Conference of the Institute of Systems,Control and Information Engineers(ISCIE), Vol.5017, 439-440, May 2004.
158.
Yuji Muramatsu, Minoru Fukumi, Norio Akamatsu and Fumiaki Takeda : Wrist EMG Pattern Recognition System by Neural Networks and Multiple Principal Component Analysis, Proceedings of the 48th Annual Conference of the Institute of Systems,Control and Information Engineers(ISCIE), Vol.5038, 481-482, May 2004.
159.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Lips extraction invariant to form change in horizontal direction, Proceedings of the 48th Annual Conference of the Institute of Systems,Control and Information Engineers(ISCIE), Vol.5018, 441-442, May 2004.
160.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Extraction accuracy and speed improvement of lips region extraction, 電子情報通信学会総合大会, Vol.D-12-60, 226, Mar. 2004.
161.
Takashi Imura, Minoru Fukumi and Norio Akamatsu : Face search by neural network based skin color threshold method, 電子情報通信学会総合大会, Vol.D-12-63, 229, Mar. 2004.
162.
Takuya Akashi, Minoru Fukumi and Norio Akamatsu : Lips Extraction for Lipreading As Human Intereface of Mobile Devices, 第46回 自動制御連合講演会, Vol.TA1-06-2, 68-69, Nov. 2003.
(Keyword)
Lips region extraction / Template matching
163.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Vowel Analysis and Classification by the Mahalanobis distance using the amplitude features, 第46回 自動制御連合講演会, Vol.FP1-09-2, 1030-1033, Nov. 2003.
(Keyword)
振幅特徴量 / マハラノビス距離 / 母音識別
164.
T Imura, Minoru Fukumi and Norio Akamatsu : Real-Time Face Area Extraction in Moving Image, 第46回 自動制御連合講演会, Vol.TA1-06-1, 66-67, Nov. 2003.
(Keyword)
moving image / real time / skin color detection / face search / neural network
165.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Facial Expression Creation using Image Warping and Neural Networks, Proc. of JACC'2003, No.FP2-02-1, 1120-1122, Nov. 2003.
(Keyword)
Face expression / Warping and Neural network
166.
Takuya Akashi, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Lips Region Extraction by Genetic Algorithms at the Moment of Speech, Proceedings of the 47th Annual Conference of the Institute of Systems,Control and Information Engineers(ISCIE), 627-628, May 2003.
167.
Hideki Matsuda, Minoru Fukumi and Norio Akamatsu : 3-Dimensional Object Recognition by an Evolutional RBF Network, Proceedings of the 47th Annual Conference of the Institute of Systems,Control and Information Engineers(ISCIE), 695-696, May 2003.
168.
Yoshitatsu Moriyama, Minoru Fukumi and Norio Akamatsu : Keyword addition to images using Interactive Genetic Algorithms, Proceedings of the 47th Annual Conference of the Institute of Systems,Control and Information Engineers(ISCIE), 633-634, May 2003.
169.
Kensuke Komobuchi, Hideaki Sato, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : The Vowel Recognition System Using the SNN, システムインテグレーション部門講演会, 341-342, Dec. 2002.
170.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Feature Analysis for EMG signals by Using an Evolutionary Method, システムインテグレーション部門講演会, 339-340, Dec. 2002.
171.
糠野 友彦, 伊藤 征嗣, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and 竹田 史章 : Developpment of a Measurement System for Food Intake based on Image Processing, システムインテグレーション部門講演会, 335-336, Dec. 2002.
172.
Kensuke Mitsukura, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Design of Genetic Face Detection and Identification System, システムインテグレーション部門講演会, 329-320, Dec. 2002.
173.
Hiroshi Nishiyama and Minoru Fukumi : Keywords Extraction form Images by Using Neural Networks, システムインテグレーション部門講演会, 327-328, Dec. 2002.
174.
Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : The EEG Detection System Uisng the Factor Analysis, 第12回 インテリジェント·システム·シンポジウム講演論文集, 413-414, Nov. 2002.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Detection Method of Face Region in Color Images by Using the Lip Detection Neural Network and the Skin Distinction Neural Network, 日本シミュレーション&ゲーミング学会秋期全国大会, 104-107, Sep. 2002.
Masayuki Shinmoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Color Image Classification Using Neural Networks, Proc. of SICE Annual Conference, Vol.TEA11-7, 1986-1990, Aug. 2002.
(Summary)
This paper presents a method for image classification by neural networks that use characteristic data extracted from images. In order to extract characteristic data, image pixels are divided by a clustering method on HSI 3-dimensional space and processed by labeling to select domains. The validity of this technique is verified by means of computer simulation.
186.
Taketsugu Nagao, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : RECOGNITION OF DRIFT ICE USING SYNTETIC APERTURE RADER IMAGES, Proc. of SICE Annual Conference, Vol.TEA11-6, 1982-1985, Aug. 2002.
(Keyword)
SAR / neural network / Self-Organizing map / false color
187.
Yuji Matsumura, Minoru Fukumi, Yasue Mitsukura, Norio Akamatsu, Yoshihiro Yamamoto and Kazuhiro Nakaura : Recognition System of EMG Patterns by Neural Networks, Proc. of SICE Annual Conference, Vol.TEA11-5, 1977-1981, Aug. 2002.
Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Toru Yamamoto : A system identification method using a hybrid-type genetic algorithm, Proc. of SICE Annual Conference, Vol.TEA11-2, 1962-1966, Aug. 2002.
(Keyword)
system indentification / bynary coded genetic algorithm / real coded genetic algorithm
189.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Skin Color Recovery from Over-Exposed Images Using Neural Networks, IEICE Technical report (NC 2002), Vol.102, No.157, 7-12, Jun. 2002.
190.
佐藤 秀明, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Emotional Speech Classification by Using Prosodic Characteristics, 電気学会全国大会, Mar. 2002.
(Keyword)
感性情報 / 感情音声 / 韻律 / ピッチ
191.
伊藤 征嗣, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Key Words Extraction for Image Search of WWW, 電気学会全国大会, 1, Mar. 2002.
(Keyword)
画像検索 / キーワード抽出 / 格子点 / neural network
192.
市原 敦子, Yasue Mitsukura, Minoru Fukumi and 安友 基勝 : A Liver Extraction Method Using the Moving Points, National Convention Record I.E.E. Japan, Vol.3, 58-59, Mar. 2002.
193.
明石 卓也, Minoru Fukumi and Norio Akamatsu : Lips Extraction with Template Matching by Genetic Algorithms, 電子情報通信学会総合大会, Vol.D-12-97, 273, Mar. 2002.
194.
長尾 剛嗣 and Minoru Fukumi : Drift Ice Recognition Using Remote Sensing DATA by Neural Netoworks, 電子情報通信学会総合大会, Vol.D-12-41, 217, Mar. 2002.
195.
松村 悠司, Minoru Fukumi, 山本 祥弘 and 中浦 一浩 : Recognition System of EMG Patterns by Neural Networks, 電子情報通信学会総合大会, Vol.D-12-40, 216, Mar. 2002.
196.
真本 昌幸 and Minoru Fukumi : Color Image Classification Using Neural Networks, 電子情報通信学会総合大会, Vol.D-12-29, 205, Mar. 2002.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : 霧発生の遺伝的予測法, システム制御·情報学会講演会, May 2000.
Et cetera, Workshop:
1.
Takuma Yoshida, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : MediaPipeを利用したハンドジェスチャーの範囲選択による文章認識, 電気学会・産業計測制御研究会, IIC-21-044, Nov. 2021.
(Keyword)
ハンドジェスチャー / 文章認識
2.
Taiga Sogame, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Discriminate H Motor Imaginations of Holding hands and Vocalization Using EEG, 電気学会・産業計測制御研究会, 55-60, Nov. 2020.
(Summary)
This paper proposes a method to discriminate motor imaginations of holding hands and vocalization. The proposed method consists of EEG measurement, EEG feature extraction, and motor imaginations classification. In EEG measurement, simple electroencephalograph is employed. In EEG feature, event-related synchronization (ERS) and event-related desynchronization (ERD) are extracted using fast Fourier transform (FFT). Support Vector Machine (SVM) is used to classify the motor imaginations of holding hands and vocation. In order to show the effectiveness of the proposed method, we conducted experiments using real EEG data.
Takeru Oda, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Development of EEG Personal Authentication System Considering the Effect of Caffeine, 電気学会・産業計測制御研究会, 41-46, Nov. 2020.
(Summary)
Changes in brain activity during caffeine intake are an issue for the spread of brain machine interface (BMI) devices to consumers. We attempt to develop an electroencephalogram (EEG) discrimination method considering the effects of caffeine. Also, this paper focuses on a personal authentication system based on EEG analysis technique. As a preliminary step to consider the influence of caffeine, this paper proposes a method to discriminate the presence or absence of caffeine and construct an authentication system. In order to show the effectiveness of the proposed method, we conducted experiments using real EEG data.
Takuma Yoshida, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Range Recognition Using Hand Gestures, 電気学会・産業計測制御研究会, 9-12, Nov. 2020.
(Summary)
This paper proposes a method to detect a range using hand gestures. The proposed method consists of hand region detection, hands recognition and gesture detection. In the hand region detection, the skin color is detected on the basis of HSV colors. In order to show the effectiveness of the proposed method, we conducted experiments with hand gestures using a web camera. The experimental results show that mean of the detection ratio for the range detection was 32.5%.
(Keyword)
AR / image processing / hand gesture
5.
Hironori Kadowaki, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Gait Recognition using Gathered Images and Deep Learning, 電気学会・産業計測制御研究会, 27-31, Nov. 2020.
(Summary)
We propose a method to recognize the gait using gathered images and deep learning. The proposed method consists of preprocessing, gathered images creation, and identification. The gathered image is created by comparing brightness values from one steps walking images. This paper employs a convolutional neural network (CNN) to extract features for gait recognition and recognize a person using them. The CNN consists of an input, three hidden, a full-connection and output layers. The hidden layers have convolutional and pooling layers. The full-connection layer has a dropout layer. Finally, we conducted experiments for gait recognition.
Shun Yamamoto, Minoru Fukumi, Momoyo Ito and Shin-ichi Ito : 空中入力数字の時系列データに対するCNNの有用性の検証, 電気学会・産業計測制御研究会, 1-4, Nov. 2019.
7.
Kazuki Shimamoto, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : 爪画像を用いた蓄積ストレス評価に関する一考察, 電気学会・産業計測制御研究会, 11-14, Nov. 2019.
8.
Ren Nozaki, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : 簡易脳波計を用いた面倒感情の検出, 電気学会・産業計測制御研究会, 23-29, Nov. 2019.
9.
Hironori Kadowaki, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : 集約画像を用いた歩容認証, 電気学会・産業計測制御研究会, 39-43, Nov. 2019.
10.
Shunsuke Takata, Momoyo Ito, Kazuhito Sato, Shin-ichi Ito and Minoru Fukumi : Evaluation of Driver Behavior Quantification using RSOM for Driver State Estimation, Conference record of 2017 Taiwan and Japan Conference on Circuits and Systems, 13, Aug. 2017.
11.
Momoyo Ito, Kazuhito Sato, Shin-ichi Ito and Minoru Fukumi : A Proposal of Suitable Driving Behavior Model Selection according to Driving Scene, Conference record of 2017 Taiwan and Japan Conference on Circuits and Systems, 15, Aug. 2017.
12.
Tomoya Tanaka, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Evaluation of the accumulated stress by expression analysis, 瀬戸内信号処理研究会 SSS2016, 14, Sep. 2016.
13.
Takashi Higasa, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : String Search in the Augmented Reality Space Using the Gesture Motion, 瀬戸内信号処理研究会 SSS2016, 13, Sep. 2016.
14.
Taiki Nonoguchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A method to Track Mosquitoes using Orientation Code Matching and Particle Filter, 瀬戸内信号処理研究会 SSS2016, 9, Sep. 2016.
15.
Taiki Nonoguchi, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Image Processing Method to Detect and Track Mosquito using Backgraound Subtraction, 瀬戸内信号処理研究会 SSS2015, 17, Sep. 2015.
16.
Shu Tamura, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : A Novel Music Database Based on Chord Pattern, Rhythm and Music Feature Extraction Method to Smilarity Evaluation, 瀬戸内信号処理研究会 SSS2015, 15, Sep. 2015.
Hikaru Shouda, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Supporting System to Grow in Description Ability for Beginners, 瀬戸内合同信号処理研究会 SSS2014, 16, Sep. 2014.
19.
Nao Tsuzuki, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Method to Detect Uncomfortable Feeling of Listeners by Biological Information - Supporting System to Hold a Conversation for Smooth Communication -, 瀬戸内合同信号処理研究会 SSS2014, 15, Sep. 2014.
20.
Shu Tamura, Shin-ichi Ito, Momoyo Ito and Minoru Fukumi : Method to Construct Database for Best Music Selection System by EEG, 瀬戸内合同信号処理研究会 SSS2014, 14, Sep. 2014.
Shin-ichi Ito, Yasue Mitsukura, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : 脳波の個人特性とユーザのエゴグラム得点との関連性, 産業計測制御研究会, 17-20, Mar. 2010.
24.
Shin-ichi Ito, Yasue Mitsukura, Katsuya SATO, Shoichiro Fujisawa and Minoru Fukumi : A Study on Pattern Classification of Left Prefrontal Cortex EEG, 第79回パターン計測部会研究会資料, 39-44, Jul. 2009.
Shin-ichi Ito, Yasue Mitsukura, 宮村 浩子, 斎藤 隆文 and Minoru Fukumi : The Proposal of the Feature Extraction Method in Weighted Principal Frequency Components Using the RGA, IEICE Technical Report, Vol.HIP2006 37, 73-78, Jul. 2006.
(Summary)
An EEG has frequency components which can describe most of the significant features. These combinations are often unique like individual human beings and yet they have underlying basic features. These frequency components are contained the important and/or not so important components, and then each importance of these frequency components is different. The real-coded genetic algorithm (: RGA) is used for selecting and weighting the principal characteristic frequency components. We attempt to construct mental change appearance model (: MCAM) of one measurement point only. In order to show the effectiveness of the proposal method, computer simulations are carried out by using real data.
Takahiro Ogawa, Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Classification of the EEG in listening the music by using multivariate analysis, IEICE Technical Report, Vol.HIP2005-33, 77-81, Jul. 2005.
Yasuyuki Takahashi, Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Morphing of Face Image by Feature Extraction, IEICE Technical Report, Vol.HIP2005-27, 43-47, Jul. 2005.
Tomohiko Nukano, Minoru Fukumi and Norio Akamatsu : Vehicle License Plate Character Recognition by Neural Networks, 第14回インテリジェント·システム·シンポジウム講演論文集, 57-60, Oct. 2004.
(Keyword)
neural network / License plate / Character recognition
58.
Ryohei Haga, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Motokatu Yasutomo : A Technique of Automatic Detection of Heart Disease, 電気学会情報処理·産業システム情報化合同研究会, Vol.IP-04-24, No.IIS-04-33, 85-90, Sep. 2004.
(Keyword)
neural network / Fuzzy reasoning / Asynergy / Left Ventricular
59.
Yoshiki Kubota, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Motokatu Yasutomo : Automatic Extraction of a Kidney Region by Using the Improved Q-learning, 電気学会情報処理·産業システム情報化合同研究会, Vol.IP-04-23, No.IIS-04-32, 79-83, Sep. 2004.
Hiroshi Kawasaki, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Music Compression System Using a Hybrid-Type Genetic Algorithm, IEICE Technical Report, Vol.NC2004-41, 77-80, Jun. 2004.
(Keyword)
Music compression / double structured genetic algorithm
62.
Ryohei Haga, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Motokatu Yasutomo : Automatic Detection of Left Ventricular Asysner Asynergy by Fuzzy Reasoning, IEICE Technical Report, Vol.NC2004-40, 71-75, Jun. 2004.
(Keyword)
neural network / Fuzzy reasoning / Asynergy / Left Ventricular
63.
Michiyo Nishioka, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Measurement of Skin Texture Using Genetic Picture Analysis, IEICE Technical Report, Vol.NC2004-39, 65-69, Jun. 2004.
Yoshiki Kubota, Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Motokatu Yasutomo : Automatic Extraction of a Kidney Region by Using the Q-learning, IEICE Technical Report, Vol.NC2004-38, 59-63, Jun. 2004.
Tomohiko Nukano, Minoru Fukumi and Norio Akamatsu : Vehicle License Plate Character Recognition in Malaysia, 第13回 インテリジェント·システム·シンポジウム講演論文集, Vol.C7-4-4, 222-225, Dec. 2003.
(Keyword)
neural network / License plate / character recognition
68.
Hideki Matsuda, Minoru Fukumi and Norio Akamatsu : 3-Dimensional Object Recognition by an Evolutional RBF Network, 第13回 インテリジェント·システム·シンポジウム講演論文集, Vol.A6-3-1, 28-31, Dec. 2003.
(Keyword)
RBF Network / real-coded GA
69.
Yuji Matsumura, Minoru Fukumi, Norio Akamatsu and Fumiaki Takeda : Wrist EMG Pattern Recognition System by Neural Networks, 第18回 ディジタル信号処理シンポジウム, Nov. 2003.
70.
Taketsugu Nagao, Minoru Fukumi and Norio Akamatsu : A Segmentation Method of Remote Sensing Images, 第18回 ディジタル信号処理シンポジウム, Nov. 2003.
71.
Yoshitatsu Moriyama, Minoru Fukumi and Norio Akamatsu : Keyword Addition to Images Using Interactive Genetic Algorithms, 第18回 ディジタル信号処理シンポジウム, Nov. 2003.
72.
Miyoko Nakano, Fumiko Yasukata, Minoru Fukumi and Norio Akamatsu : Age Estimation from Face Images Using Neural Networks, 第18回 ディジタル信号処理シンポジウム, Vol.D2-6, Nov. 2003.
73.
Yasue Mitsukura, Hiromi Bando, Kensuke Mitsukura, Minoru Fukumi and Norio Akamatsu : Finger Alphabet Recognition System Using Double Structure Neural Networks, Human Interface 2003, Vol.2232, 285-288, Sep. 2003.
Shin-ichi Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Detection of the Features of EEG Pattern Using the Factor Analysis and Neural Networks, ヒューマンインタフェース学会研究報告集, Vol.5, No.1, 127-130, Jan. 2003.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Personal Identification Using the Feature Extraction Method Based on the SPCA and the RGA, ヒューマンインタフェース学会研究報告集, Vol.5, No.1, 123-126, Jan. 2003.
(Keyword)
Personal Identification / SPCA / RGA / NN
79.
Yuuki Yazama, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : EMG Distinction Method Using a Genetic Algorithm, ヒューマンインタフェース学会研究報告集, Vol.5, No.1, 119-122, Jan. 2003.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi and Shigeru Omatsu : Keywords Extraction system in Images for Scene Analysis, ヒューマンインタフェース学会研究報告集, Vol.5, No.1, 109-112, Jan. 2003.
Hideaki Sato, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Object Detection Method Using a Double Structured Fitness Function, ヒューマンインタフェース学会研究報告集, Vol.5, No.1, 97-102, Jan. 2003.
(Keyword)
evolutionary computation / real-coded genetic algorithm / fitness function / color thresholds / image prosessing
82.
Seiki Yoshimori, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of the Objection Detection System Using, ヒューマンインタフェース学会研究報告集, Vol.5, No.1, 93-96, Jan. 2003.
Kensuke Mitsukura, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Robust Face Detection System Using the RGA, ヒューマンインタフェース学会研究報告集, Vol.5, No.1, 87-92, Jan. 2003.
(Keyword)
robust face detection / real-coded genetic algorithm / personal identification
84.
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of Face Detection System Using the NN and the Simple PCA, IEICE Technical Report, 1-6, Jan. 2003.
Yasue Mitsukura, Minoru Fukumi, Norio Akamatsu and Toru Yamamoto : A System Identification Method Using a Hybrid-Type Genetic Algorithm, 第12回 インテリジェント·システム·シンポジウム講演論文集, Vol.2-314, 451-452, Nov. 2002.
(Keyword)
System Identification / PID Control / Genetic Algorithm
Hironori Takimoto, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A Design of Face Detection System Based on the Feature Extraction Method, 第12回 インテリジェント·システム·シンポジウム講演論文集, Vol.2-303, 409-410, Nov. 2002.
Seiji Ito, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Key Words Extraction for Image Search of the Outdoor Scenes, The Papers of Technical Meeting on Industrial Instrumentation and Control, Vol.IIC-02-81, 59-63, Jul. 2002.
(Keyword)
Median Filtering / Maximin-Distance Algorithm / Integrated Small Region / Grid
89.
Hiromi Bando, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Finger Alphabet Recognition System Using Double Structure Neural Networks, The Papers of Technical Meeting on Industrial Instrumentation and Control, Vol.IIC-02-73, 15-18, Jul. 2002.
佐藤 秀明, Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : Emotional Speech Recognition with Pitch Extraction System and Neural Network, The Papers of Technical Meeting on Industrial Instrumentation and Control, Vol.IIC-02-79, 47-522, Jul. 2002.
(Keyword)
emotional speech / emotional recognition / prosodic character / fundamental frequency / neural network
91.
Yasue Mitsukura, Minoru Fukumi and Norio Akamatsu : A System Identification Method Using a Hybrid-Type Genetic Algorithm, The Papers of Technical Meeting on Industrial Instrumentation and Control, Vol.IIC-02-57, 29-32, Jul. 2002.
92.
Stephen Githinji Karungaru, Minoru Fukumi and Norio Akamatsu : Skin Color Recovery from Over-Exposed Images Using Neural Networks, IEICE Technical Report, Vol.120, No.157, 7-12, Jun. 2002.
In recent years, in connection with improvement in remote-sensing technologies using satellites that investigates the earth surface, the satellites reach far and widely and a terrestrial fine change can be observed from a remote place. In this paper, the validity of the technique by combining a neural network (NN) and a genetic algorithm (GA) is verified aiming at recognizing the drift ice with sufficient accuracy.