Adaptively Weighted 2DPCA Based on Local Feature for Face Recognition

Two dimensional principal component analysis (2DPCA) extracts the global feature of human face, but the local feature is very important to face recognition. In this paper, adaptively weighted 2DPCA based on local feature is proposed. It combines above approaches through separating original images into multi-blocks. Firstly, the face image is separated into three independent sub-blocks according to the local features. Secondly, 2DPCA is applied to the sub-blocks independently. Then the method adaptively computes the contributions made by each sub-block and endows them to the classification in order to improve the recognition performance. The experiments on the ORL and Yale face databases demonstrate the proposed methodpsilas effectiveness and feasibility.