A fast method for single image super resolution using dictionary learning

In this paper we propose a fast method for single image super resolution using self-example learning method. We first divide input image into a number of blocks. For each block a dictionary, is learnt using image patches in the block and its eight neighborhood block around it. In this learning we only use the image patches with considerable details. Each low resolution patch in image is presented as a linear combination of associated local dictionary atoms using Tikhonov regularization. In contrast to existing methods since we only use patches with high details for learning, the complexity of the proposed method is relatively low. The experimental result show the proposed method is significantly faster than existing methods whereas the performance in terms of PSNR criterions is comparable with the existing methods.