Robust Feature Based Collaborative Regression Classifier for Face Recognition

Abstract

Face recognition is a simple and effective biometric approach capable of recognizing people. Recently, linear regression classifier approach has been proposed and become popular because of its simplicity and their superior results even with low resolution face images. Later, Collaborative Representation Classifier algorithm was proposed; however, these algorithms could not provide better accuracy when images with light illumination, facial expression and face masks were used. In this paper, we propose an approach by using densely extracted SIFT features with regression classifiers. Since the dimension of features is very high, PCA is applied to reduce the dimension. Finally the performance of regression classifiers was evaluated. Experimental results show that the recognition performance of the proposed feature based approach is improved