Single Image Super-Resolution Reconstruction Using Nonlocal Low-Rank Prior

Abstract

In most practical imaging applications, the images with high resolution is desired, but most of imaging images are generally low resolution in practice, which brings many problems. In this paper, we propose an effective image super-resolution reconstruction model using nonlocal low-rank prior. Firstly, this model uses the single image as data input, and the self-similarity inside the single image is used as prior knowledges to improve the matching degree of similar image patches. Then, the reconstruction progress is modeled with maximum a posterior probability framework. Finally, a nonlocal low-rank regularization is adopted to regularize the reconstruction process, which exploits the local and global information of image to improve the reconstruction effect. Experimental results show that the proposed method has achieved better results than the existing methods.