Enhance the PCA Method to Strength Face Recognition Systems

Abstract

Information found in a human face is redundant in high rate, so that was the motivation for using PCA for face recognition is to remove redundancy, and extract the features required for comparison of faces. Difficulties with conventional PCA are Global projection suppresses local information, and it is not resilient to face illumination condition and facial expression variations. And it does not take discriminative task into account ideally.This paper propose to enhance the PCA method in face recognition systems that by, provide high accuracy of recognize faces regardless of facial parameters such as expressions, face’s angles (orientation), hairstyle and with/without classes. The enhanced PCA method concentrate on aspect that is; the images used for training are grouped into different classes and each class contains all images of a single human face with different facial parameters. Then apply PCA on each class in training database separately, so each face has a specific PCA recognize it in any frame of mind, eyes, and hair, after that correlate these PCA classes with each other to model the final total face space and projection. PCA will give a universal results face identity and it is parameters. The proposed PCA produced good results when compared with the traditional PCA recognition ratio on JAFFE database for facial expression, ORL database for face’s angles, with/without classes and hairstyle.