Abstract: An efficient nonlinear subspace learning method, kernel maximum scatter difference discriminant analysis (KMSI),was proposed for face recognition in this paper. The main idea of KMSI)is to map the input sample data into feature space by nonlinear function, and then adopt maximum scatter difference discriminant analysis(MSD) to find the solution in feature space by kernel trick. The experimental results on the Yale and ORL face image database show that the proposed KMSI)method for face recognition has higher recognition rate and more effective.

Key words: Kernel maximum scatter difference discriminant analysis(KMSD)，Subspace learning, Face recognition