Weighted module image method for face recognition

In this paper, we use the distances between the images and the average image of the within class as the weights to generate new samples. Such samples are closer to the real face than the original ones. Then merged with the module image method, we divide the weighted images to extract the feature with PCA. This method can reduce the effects of the lighting and poses before the camera, and can extract the local information of the image. It is proved having higher recognition than the Module PCA through experiments.