Research And Application Of Illumination-Robust Face Recognition

As a friendly biometric identification technology, face recognition has broad application prospects in the field of identity verification. The research on face recog-nition has become a hot topic in image processing, computer vision and pattern recognition. In recent years, face recognition has achieved satisfactory results under controlled conditions, through the efforts of researchers all over the world. However, in uncontrolled environment, face recognition is seriously challenged by variations in illumination, pose, expression and occlusion, especially, variations in illumination can significantly affect the performance of face recognition.An efficient method for illumination-robust face recognition based on sparse representation of blocks-weighted is proposed in this dissertation, which is on the basis of Sparse Representation-based Classification (SRC). The works of this disser-tation are as follows:(1) Proposed a method for face recognition based on sparse representation of blocks-weighted. On the basis of Sparse Representation-based Classification, we partition a face image into several blocks. The image partition makes much sparser representation, resulting in better performance. Furthermore, the contribution of each sub-block to face recognition can be weighted by partitioning an image into several blocks. By the experimental research and theoretical analysis, some studies are made in this dissertation, including:the relationship between effect of sparse representa-tion and flattening of dictionary matrix, the influence of partition number and down-sample size on recognition rate, the contribution of each sub-block to face recogni-tion. To validate the presented method, experiments are performed on the Yale B, Extended-Yale B and CMU-PIE databases. Experimental results show that excellent recognition rates can be achieved by the proposed method, that is, both of our recog-nition rates on the Yale B, Extended-Yale B databases are100%, and our method achieves a recognition rate of99.75%on the CMU-PIE database.(2) According to the presented method, implemented a user verification cloud system based on face recognition. The system mainly consists of two parts:the front-end and the back-end. There is a webpage for the capture of facial images in the front-end, which is developed based on the flash, and facial images can be obtained easily using the browser. Subsequently, the images are sent to the back-end, which provides cloud services for face recognition, and sends the verification results feed-back to the front-end. The test results show that the system achieves a verification success rate of95.28%in an actual environment, validating the proposed method by a practical application.