High-definition Image Restoration

In recent years, with the development of Multi-media Technology and the wide-spreaded application of Visual Communication, Image Processing has been received more and more attentions and becomes an attractive research field. In this thesis, we focus on the research of blind image restoration, which consists of the following parts:1. The image models and characters of blind image restoration are presented. Blind image deconvolution algorithms can be grouped into five classes.2. This thesis analyzes the theoretics of a space-adaptive regularization method for blind image restoration. In simulated experiment,we find some important questions on which should be paid more attention. Simulation results indicate that SAR can not only be used on blind image restoration but also on image restoration and blur identification.3. This thesis discusses and compares the nonnegativity and support constrains recursive inverse filtering (NAS-RIF) algorithm and iterative blind deconvolution (IBD) algorithm. These methods for blind image deconvolution can obtain good image estimates when free of noise. We combine more priori information of a binary image with NAS-RIF algorithm and propose a modified NAS-RIF algorithm for noisy binary images. This algorithm also considers the situation when the background is not black and overcomes the scale problem of NAS-RIF algorithm.4. A new advanced blind support-finding method is presented . This method can automatically determine the image support in blind image restoration processing and obtain good image support estimates in simulation experiments.