Multi-Pose Face Recognition Based On Sparse Representation

With the rapid development of the computer and biomedical engineering technology, the use of biometric recognition is a preferred way in ID identification. As one of the most natural and friendly biometrics, automatic facial recognition has become widely research and application. In face recognition, multi-pose problem and small sample problem are widespread, it greatly reduces the recognition rate of the original algorithm. In this paper, based on the sparse representation model, by fusing pose information of face image and using Gabor local feature to construct dictionary, according to the residual we classify the testing subjects. The paper main research work is as follows:1. The overview of the conception and basic procedure of face recognition, the methods of face recognition and the evaluation of subspace-based face recognition are also concerned. The basic theory of sparse representation is given in detail.2. Build a face recognition framework based on sparse representation, and analyze the robustness of the algorithm. Analyze problems of face recognition based on sparse representation, for example of small sample, multi-pose problem.3. To use pose information from face image and improve the recognition efficiency, a face recognition approach based on sparse representation-based classification is proposed by fusing pose information of face image, which using Gabor local feature to construct dictionary in order to enhance the robustness for pose changes. With everyone only positive image as the training sample, experimental results are tested on FERET and CMU PIE face databases, which show that the proposed method is effective and robust to the facial pose.