A Novel Method of Gradient-Based Illumination Invariant Face Recognition

The effect of illumination variation is one of challenging problems in a face recognition system, which can dramatically reduce the performance of face recognition. In order to solve this problem, this paper proposed a novel method for face recognition. First of all, illumination insensitive representation of image is extracted, which is based on the ratio of gradient amplitude to the original image intensity in the domain. Then, combined with the analysis of NS-LDA algorithm for feature extraction, and finally classified with the nearest distance classifier. Compared with the existed representative approaches, such as LBP (Local Binary Pattern, LBP), WF(Weber-face, WF),GF (Gradientfaces,GF) ,this method has better recognition rate on CAS-PEAL, Yalefaces, Extended Yale B face database.