A Combination Approach to Human Face Recognition

Abstract

AbstractIn this paper, we propose a combination approach for face recognition based on combination of twofeatures extractor schemes named Singular Value Decomposition and Gabor filters. Singular valuedecomposition (SVD) is a good method to extract image features because it has invariance for therotation and mirroring transformation, and also has better robustness for noise and light intensitytransform. Gabor filters produce perfect localization features in frequency and spatial domains. Fromtheexperimental results, the suggested approach obtains a good recognition rate. A recognition rateof more than 98% has been achieved on the ORL database.The proposed approach has also beencompared to some existing techniques and the results obtained by the proposed method are far betterthan these techniques.