A Novel Sparse Representation Classification Face Recognition Based on Deep Learning
Junying Zeng, Yikui Zhai, Junying Gan · 2015
The existing face recognition under pose and illumination variations is a challenging problem. A novel sparse recognition face recognition algorithm based on deep learning is presented in this paper. The deep learning network extracted global and local information, the deep learning network adopted the supervised Convolution restricted Boltzmann machine. The features extracted could recover the face image and reduce intraidentity variances, while maintaining discriminativeness between identities. The algorithm obtained the feature by the deep network and realized fast sparse classification by smoothed l0 norm. Experimental results on FERET face database show that the proposed algorithm can improve recognition rate and recognition speed when dealing with various conditions such as pose variation.