Analysis And Simulation Of Super-Resolution Reconstruction From Compressed Video Based On Bayesian Theory

Image resolutions acquired in real applications can not often satisfy the requirements, due to the limitations of the image acquisition systems, data compression or transmission errors. Thus, super resolution technique is presented to solve the problem, by estimating a single image of high quality and resolution from a set of low-resolution degraded observations. However, super resolution of compressed digital videos is of some difference from that of image sequences in the modeling and analysis of the system. This thesis aims at researching super resolution algorithms suitable for compressed videos based on analysis of the characteristics of the video compression techniques.There are two algorithms with good performance reviewed in the thesis: the MAP approach based on Bayesian estimation and the POCS approach based on theorem of convex sets. We utilized a modified Gunturk's MAP approach in super-resolution of real compressed videos as follows:It is necessary for super-resolution reconstruction to form an accurate...