基于稀疏表示的图像超分辨率算法研究

he purpose of image super-resolution (SR) is to reconstruction a high-resolution image from one or several low-resolution images of the same scene. Sparse-representation-based image SR is an advanced algorithm at present, but it also has shortcomings in reconstruction quality and speed. To train a dictionary with more compact structure and lower dimensions, the clustering is introduced into the joint dictionary training in this paper. Meanwhile an image reconstruction method based on adaptive block size which using different block size for reconstruction according to the images complexity is proposed. It can improve the quality of reconstruction as every block contains appropriate information for reconstruction. The results show that the proposed method can improve the quality and speed of image reconstruction.