基于深度学习和约束稀疏表达的人脸识别算法

Face Recognition via Deep Learning and Constraint Sparse Representation

目前的人脸特征匹配算法大多关注于单图像与单图像的匹配而不能有效利用图像序列之间的相关信息,因而提出了一种基于深度学习与约束稀疏表达的人脸特征匹配算法.通过CNN网络对人脸图像进行特征提取,并利用改进的稀疏表达方法自动选取相似的图像序列进行特征匹配,有效地利用了图像序列之间的相关信息.实验结果表明,该算法在LFW和AR数据库上取得了很好的效果并优于传统的SRC,L1-norm和CRC-RLS算法.

The matching between single image and single image has attracted much attention in the current face feature matching algorithms,in order to make use of the correlation information between image sequences effectively, a face feature matching algorithm based on deep learning and constraint sparse representation was proposed. The feature extraction of face images was carried out through CNN network, and an improved sparse expression method was used to automatically select similar image sequences for feature matching, so that the correlation information between image sequences could be effectively utilized. Experimental results show that the proposed algorithm can achieve better result in LFW and AR databases, and is superior to other face feature matching methods,such as SRC,L1-norm and CRC-RLS.