Face Recognition Using Feature Optimization and Ν-support Vector Learning

A new face recognition system is introduced and analyzed in this paper. The system utilizes a novel statistical pattern recognition method to optimize the feature selection process. The optimized feature set feeds a classification module, which is based on the modified ν-support vector machine approach (ν-SVM). The optimized feature set reduces the burden of the subsequent ν-SVM classifier and improves its learning speed and classification accuracy. The paper includes, simulation studies and comparative evaluation with several existing systems on the ORL face database. Results indicate that the proposed system has excellent performance achieving the lowest error rate reported to date for the ORL face database using only a very small set of features.