Face Recognition System Based on Gabor Wavelets Transform, Principal Component Analysis and Support Vector Machine

Abstract

Face Recognition is a well-known image analysis application in the branches of pattern recognition and computer vision. It utilizes the uniqueness of human facial characteristics for personnel identification and verification. For a long time, the recognition of facial expressions by using computer-based applications has been an active area of study to recognize face scheme through a face image database. It is used in a variety of essential fields of modern life such as security systems, criminal identification, video retrieval, passport and credit cards. In general, face recognition process can be summarized in three distinct steps: preprocessing, feature extraction, and classification. At first, histogram equalization and median filter are applied as preprocessing methods. Secondly, Gabor wavelets transform extracts the features of desirable facial characterized by, orientation selectivity, spatial locality, and spatial frequency to keep up the variations caused by the varying of facial expression and illumination. In addition to that, Principal Component Analysis methodology (PCA) is used in dimensionality reduction. At last, Support vector machine (SVM) is applied in classifying the feature of the image according to the classis of every mage. In order to test the approach used in this research, experiments were running on Yale database of 165 images from 15 individuals in MATLAB environment. The results obtained from the experiments confirmed the accuracy and robustness of the proposed system.