A Selective Ensemble Face Recognition Algorithm Based on Multiscale Patch Collaborative Representation

In order to further improve the recognition performance of face recognition system under the condition of small sample size, we propose a novel selective ensemble face recognition algorithm based on multiscale patch collaborative representation.Firstly, the newly proposed method has the ability to identify key characteristic subblocks through total variation local patch weighted. Secondly, it also effectively improves the generalization ability and robustness of the ensemble classifier through selective ensemble of multiscale patch collaborative representation.Numerical experiments were carried out on three standard face databases under different acquisition environments with a variety of light, facial expression and posture changes. The experimental results indicate that the newly proposed algorithm has better recognition performance and robustness than the existing state of the art sparse representation classification algorithm.