Research On Based On Singular Value Decomposition And KL Projection Fusion Decision

Human face recognition is a very difficult and important task in the field for pattern recognition, which has a wide field of potential applications. It is a kind of better biological recognition technique. Computer Face Recognition is one of the most challenging tasks for computer vision and pattern recognition. The most useful applications contain bank system, military security, security checking, etc. In this paper, research of the methods based on algebra feature, and the Singular Value Decomposition and KL Projection Fusion Decision is presented, which bases on the Principal component Analysis. This algorithm extracts the principal features by the techniques of the simulative K-L transform, then get the features vectors by the Singular Value Decomposition, and the principal component analysis and the Fisher linear discriminate analysis. Finally fusion decision is adopted to get the final result of face recognition based K-L projection and singular value. This fusion result would be saved and used to decision. This algorithm not only has advanced correct recognition rate feature that K-L projection algorithm has but also this algorithm represented a feature that Singular Value Decomposition have be possessed of independent of translation, move and rotation to describe image sample. This organic Fusion Decision eliminated influence that image intervolve correlativity descent had made when face pose in face library have changed .It had be more advance correct recognition rate. Experimental results show the feature vector based on the Singular Value Decomposition and K-L Projection Fusion Decision has good stability , discriminate ability and recognition rate than singular K-L Projection Algorithm and Singular Value Decomposition Algorithm.