A Novel Subspace Discriminant Locality Preserving Projections for Face Recognition

Abstract

This paper addresses Small Sample Size (3S) problem of Locality Preserving Projection (LPP) approach in face recognition. It is well-known that the dimension of pattern vector obtained by vectorizing a facial image is very high and usually greater than the number of training samples. Under this situation, 3S problem always occurs and direct utilizing LPP algorithm is infeasible. To deal with this limitation, a novel subspace discriminant LPP approach (SDLPP) is proposed in this paper based on modified LPP criterion and supervised graph. Furthermore, our SDLPP approach has low computational complexity. Two face databases, namely ORL and FERET databases, are selected for evaluations. Compared with some existing sate-of-the-art LPP based methods, the proposed SDLPP method gives the best performance.