Super-Resolution Restoration Of Video Sequences Image

Super-resolution restoration techniques of video image is of great value and broad prospects in the video surveillance, satellite images, and other fields. When the CCD camera is sampling for object which has the rich features in the spatial frequency, because the CCD pixel size is limited, we have to get low-resolution images. Super-resolution image restoration techniques can be used in the video frame image sequence between the redundant information, reconstruction and the elimination of surplus lower-resolution images mixing effect. This paper focus on motion estimation algorithm and interpolation algorithm which is the two key issues of restoring of super-resolution video sequence, and have a more in-depth study, we restorate a high-resolution image from Under-Sampling image sequence. After analysis of the traditional theory and methods, we present the new algorithm:(1) Motion estimation is described the changes of model parameters as the time, it is very important in the super-resolution restoration of multi-frame images and video sequences. The video image sequence is improved spatial resolution depends largely on the accuracy of motion estimation operator. the paper studies and compares the various motion estimation method, Propose a registration algorithm based on time-domain and frequency domain.(2) By motion estimation, we can get the corresponding relationships between image points. It is necessary to carry out precise image sequence registration. In the super-resolution algorithm, we improve the traditional bicubic interpolation, and get adaptive bicubic interpolation. We get a high-resolution image from several low-resolution images, we have very good results.