Sub-pattern based Maximum Margin Criterion for face Recognition

In this paper, a novel supervised feature extraction method called Sub-pattern based Maximum Margin Criterion (SpMMC) is developed for face Recognition. Unlike Maximum Margin Criterion (MMC) method which directly extracts the global features from the whole face image, the proposed SpMMC method separately extracts the local features from the sub-images partitioned from the original face image. Moreover, in order to take the correlation of different sub-images into account, the configural structure of sub-images from the same face image is considered in the proposed method. Finally, experimental results on three standard face databases (Extended YaleB, CMU PIE and AR) demonstrate that the proposed SpMMC outperforms some other state-of the-art sub-pattern based face recognition algorithms.