The New Theory of Discriminating and Generating Optimal-face

In this paper, a new theory of optimal-face generating is proposed to improve recognition rate of face image. Its aim is to solve how to select a relatively better face image to be recognized in dynamic face recognition system. The proposed theory improves the system performance by integrating the effects of multiple factors including pose, illuminating, expression and resolution. A new system is built between face detecting system and face recognition system which is called the optimal-face generating system. The method of optimal-face generating is to generate a face image suit for recognition through searching for or synthesizing or transforming images. The optimal-face is the one which has frontal view or small rotations, neutral expression, even illumination and high resolution. The recognition rate in dynamic recognition system is greatly improved after applied the proposed theory and new system