POSE, ILLUMINATION AND EXPRESSION INVARIANT FACE RECOGNITION USING 2D-PRINC'IPAL COMPONENT ANALYSIS

Abstract

In recent _times it becomes easier to detect and identify faces from still images. The detection of faces using computational approaches is a quite challenging task. It is evident that automatic face recognition system is sophisticated and reliable means to fulfil the security norms which is an indispensible criterion regarding this. Moreover, in present time face recognition is associated to the prime vicinity which has a direct relevance to computer vision technology. It is an ongoing field of research which has really vast application pertaining to criminal identification and authentication; also it has given secured system to prevent most of the cyber crimes. Being a reliable technique face recognition technology is still facing challenges like pose, illumination and expressions problem. Person identification can be done through various biometric characteristics. Out of other biometric characteristics face is utilized for identification of an individual's, because face is unique biometric characteristics. Algorithm has been devised using Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA) and 2D-Principal Component Analysis (2D-PCA). Here PCA and 2D-PCA are compared to boost up the performance in terms of recognition accuracy. Approaches are devised to frame face detection algorithms to detect faces in color images. Although in recent time face recognition is made feasible with conventional tools like statistical, neural network and hybrid approaches. In this thesis emphasis has been made on application of present time algorithms to improve the computation performance and achieve a high level of accuracy. In this work databases from Olivetti and Oracle Research Laboratory (ORL) and Yale are incorporated with applications of mentioned computational techniques to develop the required theme