Based On The Sequence Of Mixed Map / Pocs Super-resolution Image Reconstruction Algorithm

As people's requirements to image resolution improved, image super-resolution reconstruction technique came into being. Image super-resolution reconstruction technique reconstructs a high-resolution image from low-resolution images which are degraded by warping, blurring and noising. The technique is applied to public safety field, digital TV domain, military in remote sensing, medical field, and so on, which has become one of the hottest topic in current image processing area.Generally speaking, image super-resolution reconstruction technique includes image registration, interpolation, de-blurring and de-noising. Based on the research of two popular spatial domain algorithms (POCS and MAP) of super-resolution reconstruction, hybrid MAP/POCS algorithm is also researched. The main research work is as following:Firstly the applications, research dynamic and development direction at home and abroad of image super-resolution reconstruction are introduced, and then two important aspects (image registration and reconstruction algorithm) in the process of image super-resolution reconstruction are researched. Several common image registration algorithms are introduced, especially the keren sub-pixel image registration algorithm which is combined with the existing spatial domain reconstruction algorithm together, and applied in sequence image(obtained by video) super-resolution reconstruction.Secondly two popular spatial domain algorithms (POCS and MAP) of super-resolution reconstruction are studied in-depth, and the advantages and disadvantages of them are compared. The existing hybrid MAP/POCS reconstructions algorithm are also studied.Then the complementarity of characteristics of POCS algorithm and MAP algorithm can inhibit each other's shortcomings to some extent, but the shortcomings of hybrid MAP/POCS algorithm are still the existing. Thus a new improved hybrid MAP/POCS algorithm is proposed in the text. In allusion to edges and details retention ability of hybrid MAP/POCS, the edges of interpolated images are revised before reconstruction. The Laplace-gauss operator is used for constraint high- resolution image.Finally POCS, MAP, hybrid MAP/POCS and improved hybrid MAP/POCS are programmed. Multi-group images are used to verify the effectiveness of the improved algorithm in the text. Experimental results show that both the simulated degraded images, and images obtained from real video, can be reconstructed effectively by the improved algorithm.