Face image super-resolution through POCS and residue compensation

In this paper, we propose a novel face image super-resolution using the method of projection onto convex sets (POCS) and residue compensation. First, the initial reconstructed high-resolution image, which is similar to the original high-resolution image, is built under the POCS algorithm. Second, the residue compensation is estimated. The higher frequency information is reconstructed by learning two corresponding sets of high- and low- resolution training residue images. The optimal super-resolution face image, which has more image content, is hallucinated by compensating the residue to the initial reconstructed high-resolution image. Experiments demonstrate that our approach does render high quality super-resolution faces.