SSIM-Based Sparse Image Super-Resolution with Rotation Strategy and Nonlocal Regularization
Caiyan Wang, Juan Li, Jin Chu Wu, Jin Liu · 2022
The methods based on sparse representation have been proved to be effective in solving the problem of image super-resolution. In this paper, we propose a structural similarity (SSIM) based sparse image super-resolution method. We incorporate SSIM index, a perceptual image quality assessment measure, into the sparse representation model. The modified orthogonal matching pursuit (IOMP) algorithm is used to calculate the sparse coefficients. At the same time, we use rotation strategy for sparse reconstruction from multiple angles. Moreover, we further regularize the sparse reconstructed image using the global reconstruction constraint combined with nonlocal similarity. Our experimental results demonstrate the effectiveness of the proposed method in both objective and perceptual quality.