摘要： 为充分利用样本的类别信息，提出一种有监督的稀疏保持近邻嵌入算法(SSNPE)。该算法结合稀疏表示和保持近邻的思想，根据先验类标签信息保持局部邻域的固有几何关系。采用最小近邻分类器估算识别率，测试结果表明，在姿态、光照和表情变化的情况下， SSNPE都具有较高的识别率。

关键词: 人脸识别, 稀疏表示, 保持近邻嵌入, 有监督, 稀疏重构权值

Abstract: In order to make full use of the classification information of samples, an Supervised Sparsity Neighborhood Preserving Embedding (SSNPE) algorithm is proposed. It combines the ideas of Sparse representation and NPE, so it can hold the strong discriminating power while preserving the intrinsic geometry relations of the local neighborhoods according to prior class-label information. Nearest neighborhood algorithm is used to construct classifiers, the proposed method is tested and evaluated in the Yale face database and AR face database. Experimental results show that SSNPE has good performance even if pose, illumination, face expression change.

Key words: face recognition, sparse representation, Neighborhood Preserving Embedding(NPE), supervised, sparse reconstruction weight

中图分类号: 

• TP391.41