篇名 | Manifold-Regularization Super-Resolution Image Reconstruction |
---|---|
卷期 | 28:1 |
作者 | Xian-Hua Zeng 、 Su-Li Hou |
頁次 | 119-136 |
關鍵字 | medical imaging 、 nonlocal self-similarity 、 sparse representation 、 super-resolution 、 manifold smoothing 、 EI 、 MEDLINE 、 Scopus |
出刊日期 | 201702 |
DOI | 10.3966/199115592017022801010 |
The conventional sparse coding-based super-resolution image reconstruction methods cannot preserve the image local smoothing structures, i.e., has not considered the smoothness between patches in the reconstruction process. In this paper, we introduce the manifold regularization to constrain the image patches lay on the intrinsic smoothing manifold, and incorporate the image nonlocal self-similarity into sparse representation model to improve the accuracy of the sparse coefficients, thus more accurate high-resolution patches can be obtained and used to reconstruct the better high-resolution image. Accordingly a novel Manifold-Regularization Super- Resolution Image Reconstruction algorithm (MSIR algorithm) is proposed. Experimental results on benchmark medical images and the publicly available medical image sets (100 images) demonstrate the effectiveness of MSIR algorithm outperform the four classical super-resolution reconstruction methods.