Generating Optimal Face Image in Face Recognition System

In this paper, a new method of generating optimal face image is proposed for improving the performance of automatic face recognition system. In order to overcome the influences of variations in face pose and expression, it is essential to generate optimal face image for the face to be recognized. The optimal face image is the one which has the minimum pose rotation and neutral expression in image sequences. Through estimating face pose and expression and combining tensors decomposing in multi-linear algebra, we can generate optimal face images from three-route video inputs of CCD in parallel detecting system. A new module of optimal face image generation is built between face detecting module and face recognition module. The proposed method overcomes the problem which large pose angles affect face recognition rate. The recognition rate is improved after applying the proposed method.