Regularized Large Margin Classifier

In this paper, a novel linear projection classification technique, termed regularized large margin classifier (RIMC), is developed in this paper. Through adding a regularized term (within-class scatter) to the margin criterion of the classical large margin classifier, the margin between two classes of the resulting projection feature vectors achieves maximized while the within-class scatter can achieve minimum. Finally, the proposed classifier is tested on ORL face database and AR face database. The experimental results verify the effectiveness of the proposed method.