Abstract: This paper deals with combined approach of principle component analysis （PCA） and independent component analysis （ICA） to get the representation basis of face image set and proposes a new multi-step approach to extend support vector machine （SVM） capability to deal with multi-class face recognition by incorporating with the elimination strategy. Based on the one-against-one classifying strategy,it sorts the discrimination functions according to their own Vapnik-Chervonenkis confidence and uses the redundancy among them to decrease the discrimination error in case of rejecting decision. Experiments with two face-databases show that the proposed method has reached a higher recognition rate with reasonable time cost.