Enhanced (PC)A for Face Recognition with One Training Image per Person

Recently, a method called (PC)A was proposed to deal with face recognition with one training image per person. As an extension of the standard eigenface technique, (PC)A combines the original face image with its first-order projection and then performs principal component analysis (PCA) on the enriched version of the image. It was reported that (PC)A could achieve higher accuracy than the eigenface technique through using 10%-15% fewer eigenfaces. In this paper, we generalize and further enhance (PC)A along two directions. In the first direction, we combine the original image with its second-order projection as well as its first-order projection in order to acquire more information, and then apply PCA on the derived images. In the second direction, instead of combining the original image with its projections, we regard the projections of the original images as derived images that could augment training information, and then apply PCA on all the training images available, including the original ones and the derived ones. Experiments on the well-known FERET database show that the enhanced versions of (PC)A are about 1.6% to 3.5% more accurate and use about 47.5% to 64.8% fewer eigenfaces than (PC)A.