Robust face recognition by fusion local singular value feature and deformable model

Presently, face recognition has two main barriers, which are the variation of illumination, expression, pose and the occlusion and disguise respectively. The problem of robust identification human faces with varying expression and illumination, as well as occlusion and disguise will be researched in this paper. Firstly, singular value decomposition will be used for the face image, casting the singular value vector of test face image as a linear combination of the singular value vectors of face database and used deformable model representation; then, match optimization deformable model for solving combinatorial coefficient; finally, according to the sparse nature of coefficients for classification, and use slice-weighted strategy to further improve the robustness. Experimental results on Extended Yale B Database and AR Database shows that this method is very effective for face recognition and can significantly improve the robustness and stability of disguise and occlusion.