Face Recognition Metgods Based On Inverse Spectral, Dimension Reduction And Classification

Face recognition has wide applications. Thus it is still a challenging issue to develop an effective and feasible recognition method. In this thesis, I have studied robust methods for face recognition. The purpose is to make the methods have a high recognition rate when there exist blur images in the face database.This paper discusses and conducts experiments about the existing several methods of evaluating the image quality. They include inverse spectral assessment method, wavelet transform method, gradient profile sharpness histogram method, adjacent pixel difference method, cumulative probability blur detection method and faces quality assessment method. In comparison with these methods, this paper makes improvements to the inverse spectral assessment method. Moreover, we discuss ensemble classification methods which can be applied to the image separation. Experimental results demonstrate the effectiveness of the proposed approach. Finally, we discuss the face recognition methods based on feature selection.