Optimized Multi-task Sparse Representation based Classification Method for Robust Face Recognition
Sparse representation based classification(SRC)method has become a hot topic in recent years.To address the alignment problem,feature-based(such as SIFT)SRC method has been proposed.But it does not always work ideally for the contiguously occluded face images,which may due to the issue of features reliability for multi-task recognition.Based on theoretical analysis,an optimized multi-task feature-based SRC method is proposed by evaluating features representation reliability(RR).In this paper,firstly,we introduce the principle of the multi-keypoint descriptor based SRC(MKD-SRC)method.Then we present the main thought of designing the formula of RR,with which an optimized feature-based SRC method is proposed.Finally,experiments on Yale and AR database are performed.The performance of the proposed method is compared with those of the methods of MKD-SRC,SIFT-matching and original SRC.Experimental results show that the proposed method is more robust for simultaneously handling the scenarios of un-alignment,occlusion,expression or illumination variations.
Representation reliability evaluation sparse representation sparsity optimized sparse representation based classification method face recognition
Bo Sun Feng Xu Dongyang Liu Qi Kuang Jun He
College of Information Science and Technology Beijing Normal University Beijing, China
国际会议
厦门
英文
813-818
2014-08-19（万方平台首次上网日期，不代表论文的发表时间）