An adaptive single image method for super resolution

In this paper we propose an adaptive method for single image super resolution by exploiting the self-similarity. By using similarity between patches of input image and a down sampled version of the input image, we create super-resolution image. In the proposed method, first we segment input image. For each segment if variance of intensity is significant, we increase overlap between patches and reduce the patch size. On the contrary, for image segments with low detail we decrease the overlap between patches and increase the patch size. The experimental result showed, the proposed method is significantly faster than the existing methods whereas the performance in terms of PSNR criterions is comparable with the existing methods.