Satellite Image Super-Resolution Using Overlapping Blocks via Sparse Representation

In image processing the Super-Resolution (SR) has played an important role by acquiring High-Resolution (HR) images from the corresponding Low-Resolution (LR) images. In this paper, a Super-Resolution technique for satellite images is proposed but it can be used on images of different nature. In the current proposal to achieve a HR image, it is necessary an intermediate step, which consists in performing an initial interpolation, then features are extracted from this initial image, here, it is necessary to reduce the information obtained by the features extraction via principal component analysis (PCA). Patches are extracted from the initial image and the reduction via PCA. For each patch, the sparse representation is obtained and then, it is used to recover the HR image. By using the quality objective criteria PSNR and SSIM, the proposed technique is evaluated and shows a superiority in comparison against other existing proposals.