Linear Discriminant Analysis in Complementary Subspace for Face Recognition

Based on random subspace,a complementary subspace linear discriminant analysis (LDA) approach is presented for face recognition.Compared with the Fisherface and the null space LDA which only perform the discriminant analysis in the principal and null subspaces respectively,the proposed method extracts discriminative information from the two subspaces simultaneously and combines the two parts discriminative features on the feature level.Furthermore,random subspace is generated under the most suitable situation for the null space and all random subspaces are integrated on the decision level.Experiments demonstrate that the proposed method can effectively solve the small sample size problem of LDA.