A Simplified FDDL Algorithm for Face Recognition

Abstract

Abstract. In this paper, a modified Fisher Discrimination Dictionary Learning method is introduced, which is suitable for face recognition without and with disguise. In the proposed method the new optimization model is establish, so as to preserve intrinsic complementary penalty as well as able to simplify calculation of original algorithm. The effective and efficient of the proposed method is demonstrated experimentally by considering the YALE, ORL, and AR face database.