Research On Collaborative Representation Of Single Sample Face Recognition Based On Dictionary Learning

Face recognition has always been a research hot topic in the field of computer vision.Compared to other biometrics,the face recognition has friendly,easy-to-access,non-contact features,making it a wide-range of applications.Although the face recognition technology is relatively mature at this stage,single-sample face recognition is still a challenging task.In this paper,we mainly study the collaborative representation related single-sample face recognition algorithm based on dictionary learning.(1)An face recognition algorithm of diversity and extended patch collaborative representation for a single sample per person is proposed(DEPCRC).Considering the fact that patch-based method can effectively avoid the effect of Intra-class variations,and the fact that auxiliary dictionary learned from a generic training set can collaboratively represent the facial Intra-class variations,thus,we extend patch collaborative representation based classification into the single-sample scenarios by using the auxiliary dictionary and learn patch weight by exploiting regularized margin distribution optimization.For more samples information,we construct diversity training samples by the means of cropping,adding noise and gray scale transformation.Testing in the Carnegie Mellon University posture illumination face(CMU-PIE),Extended Yale B,Aleix Martinez and Robert Benavente(AR)and Labeled faces in the wild(LFW)face datasets,The experimental results show that DEPCRC has a high recognition rate under the condition of a single sample.(2)An face recognition algorithm of auxiliary dictionary of diversity learning for face recognition with a single sample per person(ADDL)is proposed.We first produce virtual face images by mirror images,square block occlusion and grey transform,and then learn auxiliary dictionary of diversity by a designed objective function.Considering patch-based method can reduce the effect of facial Intra-class variations,the face image is processed by patch processing,and the extended collaborative representation classification algorithm is used to identify each patch image,and finally,synthesize the recognition results of all the patches of each face image as the output.The experimental results show that DEPCRC has a high recognition rate under the condition of a single sample.