Study On Super-resolution Reconstruction Algorithm

Image restoration technique is widely used. Image super-resolution reconstruction of image processing in recent years has been a research hotspot in the field of image restoration. It has a researching significance in theoretically as well as a broad application in practically. Super-resolution restoration is an economic image processing method that restores higher resolution images containing more details from an image sequence of lower resolution, through such methods as image fusion, de-blurring and de-noising. Image super-resolution reconstruction breaks the restrictions of the first generation of the traditional image restoration technology. And with the strict of theory and extensive application, it has made a large number of experts, researchers and scholars to participate in and explore. Lots of foreign research institutions also have made it to be their important direction of study.This paper has a comprehensive analysis of current domestic and international image super-resolution reconstruction technique research. Aiming at the restoration of degraded image sequence super-resolution reconstruction, image sequence degraded process and reconstruction process of ill-posed problem in super-resolution reconstruction are Analyzed in this paper. Use the method of least square method, adaptive regularization and distribution iteration to reconstruct Image sequence. The first iteration is from the single low-resolution image to the interim image to avoid computational complexity of sports fuzzy matrix. The second iteration greatly reduces the number of iteration with highly relevant images. At the some time, save memory space what the Iteration required and exerts the advantages of both. Also the introduction of adaptive regularization method is used, so that image reconstruction can be better.The experimental results show that this method is effective, and the PSNR of the reconstruction image at peak and the subjective vision effect were significantly improved.