Multi-pose Face Recognition Algorithm Based on Sparse Representation

Abstract

Face recognition is a key problem in computer vision, however the performance of multi-pose face recognition has not satisfied by us. In this paper, we proposed a novel sparse representation based multi-pose face recognition algorithm. The sparse representation of a signal refers to a linear combination of several elements of a specific dictionary, and then we covert an optimization problem to the sparse representation solving. Next, the test sample is represented as an overcomplete dictionary, in which the elements refer to the training examples. Then, the multi-pose face recognition problem is solved by classifying the testing image to a suitable class by solving an optimization problem. Finally, to test the effectiveness of the proposed algorithm, FRGC 2.0 database are utilized to testify the performance of our method and other related method. Experimental results prove that our method can achieve higher accuracy for multi-pose face recognition than LLR and DBN.