Research on KPCA and NS-LDA Combined Face Recognition

Abstract

Kernel Principal Component Analysis (KPCA) is the promotion of PCA in kernel space, Null space LDA can be directly employed to choose a set of optimal projection vectors by preserving effective information of null space of within-class scatter maximizing ratio of the between-class scatter to the within-class scatter. This paper puts forward the method about KPCA plus NS-LDA for feature extraction and is applied in face recognition study, it enhances face recognition performance by virtue of combining the advantages of KPCA makes use of data high order characteristic and good divisibility of NS-LDA projection matrix. the experimental results show this method could effectively improve the recognition rate.