Medical Image Super-resolution Based On Sparse Representation

In the medical imaging process, image degradation occurs due to the poor imaging environment, the limitations of physical imaging systems as well as quality-limiting factors such as noise and blur. A solution to this problem is the use of super-resolution(SR) techniques. Super-resolution has great practical and theoretical value in medical image processing. Sparse representation based Super-resolution recently becomes the focus of much interest. Aiming on increasing the resolution of medical image, in this paper, we study sparse representation based super-resolution algorithms.In this paper, super-resolution is used as a post-processing of medical image. Firstly, based on spatial smooth subspace learning model, a smooth regularization term is introduced into the principal component analysis algorithm so that the trained dictionary can be more precise. Secondly, through analysis of the relationship between noise and the nonlocal similarity search window, an adaptive nonlocal similarity search method is proposed, with which the noise robustness of the super-resolution algorithm is improved. Experiments show that the images restored by the proposed method have shaper edges and less noise than many other algorithms.In this paper, super-resolution is used to improve medical image quality. This study not only provides a reference for medical diagnosis but also provides a reference for application of super-resolution in medical imaging.