Stephen Karungaru

Face recognition using genetic algorithm based template matching

We present a face recognition method using template matching. Template matching is performed with the help of a genetic algorithm to automatically test several positions around the target and also adjust the size of the template as the matching process progresses. We use two kinds of templates for each face. One is based on edge detection and the other depends on the YIQ color information from the face. Our template is a T-shaped region symmetrical between the eyes, covering both eyes, the nose and mouth. Our features of interest to achieve face recognition are therefore the eyes, nose and the mouth. We ignore the shape of the face so as to have a small template for faster matching and also because the effect of the shape does not result in a significant increase in the overall final accuracy. We conducted a simulation experiment to verify our idea and also did a comparative experiment using a distance measure face recognition method.

Face recognition algorithm using wavelet decomposition and Support Vector Machines

Face recognition algorithm is a very promising technique in biometric authentication. However, the recognition precision can be affected by many factors, such as feature extraction method and classifier selection. In this paper, a novel algorithm for face recognition is presented according to the advances of the wavelet decomposition technique and the Support Vector Machines (SVM) model. The extracted features from human images by wavelet decomposition are less sensitive to facial expression variation. As a classifier, SVM provides high generation performance without transcendental knowledge. First, we detect the face region using an improved AdaBoost algorithm. Second, we extract the appropriate features of the face by wavelet decomposition, and compose the face feature vectors as input to SVM. Third, we train the SVM model by the face feature vectors, and then use the trained SVM model to classify the human face. In the training process, three different kernel functions are adopted: Radial basis function, Polynomial and Linear kernel function. Finally, we present a face recognition system that can achieve high recognition precision and fast recognition speed in practice. Experimental results indicate that the proposed method can achieve recognition precision of 96.78 percent based on 96 persons in Ren-FEdb database that is higher than other approaches.

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

Improving mobility for blind persons using video sunglasses

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.

Recognition of Wrist Motion Pattern by EMG

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

Morphing face images using automatically specified features

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

Wrist EMG signals identification using neural network

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.

Feature extraction for face detection and recognition

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.

Classification of fingerprint images into individual classes using Neural Networks

In this paper, we propose a classification system for fingerprint images that is based on the number of registered fingerprint persons. Most automated fingerprint identification systems use prior classification of fingerprint for improvement of efficiency verification using minutiae as features. However, methods that use fingerprint minutiae needs improvement because they are limited to the number of classable data. Therefore, many fingerprints are classified together, consequently taking a long time to match and verify a given fingerprint. In this work, we propose a system that classifies fingerprint patterns into individual classes. Instead of the classification using minutiae, we propose a classification system that is based on individual features and the number of registered persons. Efficiency verification improves because we donpsilat need to compare an input fingerprint image to all registered fingerprint images using this system. The proposed system carries out classification using neural network.

A GA-ACO hybrid algorithm for the multi-UAV mission planning problem

Multi-UAV mission planning is a combinational optimization problem, that aims at planning a set of paths for UAVs to visit targets in order to collect the maximum surveillance benefits, while satisfying some constraints. In this paper, a genetic algorithm and ant colony optimization hybrid algorithm is proposed to solve the multi-UAV mission planning. The basic idea of the proposed hybrid algorithm is replacing the bad individuals of the GAs population by new individuals constructed by ant colony algorithm. Also, an efficient recombination operator called path relinking is used for mating. A population partition strategy is adopted for improving the evolving efficiency. Experimental results suggested that the proposed hybrid algorithm can solve the test instances effectively in a reasonable time. The comparison study with several existing algorithms shows that the proposed algorithm is competitive and promising.