Yasue Mitsukura

Recognition of EMG signal patterns by neural networks

The 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 neural network learns FFT spectra to classify them. Moreover, we perform the principal component analysis using the simple principal component analysis before we perform recognition experiments. It is shown that our approach is effective to classify the EMG signals by means of computer simulations.

License plate detection system by using threshold function and improved template matching method

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.

License Plate Detection Using Hereditary Threshold Determine Method

License plate recognition is very important in an automobile society. Also in it, since plate detection has big influence on subsequent number recognition, it is very important. 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. 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.

Feature analysis for the EMG signals based on the class distance

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.

Design and evaluation of neural networks for coin recognition by using GA and SA

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.

Face detection and emotional extraction system using double structure neural network

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). In conventional methods, if there is 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 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. First 100 lip color, 100 skin color and 100 background pictures, which are changed into 10 /spl times/ 10 pixel, are prepared for training. The validity was verified by testing images containing several faces.

Feature Generation by Simple-FLDA for Pattern Recognition

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). In a field of pattern recognition or signal processing, the principal component analysis (PCA) is popular for data compression and feature extraction. Furthermore, iterative learning algorithms for obtaining eigenvectors in PCA have been presented in such fields, including neural networks. Their effectiveness has been demonstrated in many applications. However, 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

A feature extraction of the EEG during listening to the music using the factor analysis and neural networks

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.

A system identification method using a hybrid-type genetic algorithm

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.

Face detection and emotional extraction system using double structure neural networks

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/spl times/10 pixels, are prepared for training. The validity was verified by testing images containing several faces.