Revised NMF with LDA based color face recognition

Color face is composed of information of different channels and provides more cues for recognition than grey scale one. A revised NMF with LDA based color face recognition approach is proposed in this paper. Block diagonal matrix is exploited to encode color information of different channels. And block diagonal constraint is imposed on NMF to factorize matrix. Class information of face samples is integrated into the approach through LDA on factorization coefficients. Experimental results on CVL and CMU PIE databases verify the effectiveness of the proposed approach.