关键词: 稀疏表示, 分类问题, 自步学习, 加权系数, 人脸识别

Abstract: In recent years, a sparse representation based classification (SRC) has become a hot issue which has been great successful in face recognition. However, the sparse coding may reconstruct a test sample by training samples which are far from the test sample. Meanwhile, SRC tends to lose locality information and thus produce unstable classification results. In this paper, a self-paced learning weight sparse representation based classification (SPL-WSRC) has been proposed. It can effectively eliminate the training samples which are far from the test sample, so as to avoid reconstructing a test sample by these training samples. In addition, the weight coding is effective for locality information, so the proposed SPL-WSRC can improve the classification accuracy and stability. The experimental results on the three classical face databases show that the proposed SPL-WSRC algorithm is better than the original SRC algorithm. The effect is more obvious, especially when the training samples are enough.

Key words: sparse representation, classification, self-paced learning, weighting coefficient, face recognition