Research Of Super-resolution Restoration Based On Image Registration

With the rapid development of digital imaging technology, digital imaging equipments have been widely used. However, in some situations, the digital image and video can not met the demands because of the limited resolution. This thesis studies the digital image super-resolution restoration technology to generate high resolution images or video sequence. Super-resolution restoration is an important embranchment of image fusion,and the use of this technique becomes more and more widely and the request of it becomes higher and higher in remote sensing, video frequency, iatrology, public security and so on.The development of super-resolution restoration is described in terms of their application background, used methods and future research directions respectively. Also, a high-resolution restoration algorithm for images with large comparatively movement is proposed. This algorithm based on the basic of image registration, for the redundancy high-resolution component between the low-resolution frames, the high frequency was distilled through the local MAP estimation. By compensating the high-frequency component, the high-resolution images are recovered. This algorithm solved the problems in traditional methods that the translation and rotation should not be too large. The result of experiments proved that the proposed algorithm outperforms the other throughout the sequence, and this algorithm has lower computational cost than the alternatives.