The Research On Superresolution Reconstruction Of A Sequence Of Images About The Same Scene

Multi-frame image super-resolution reconstruction technique refers to obtain a single high resolution image obtained from several low resolution images which must be of the same scene and must be taken from slightly different angles. This technique doesn't need to change the existing imaging equipment, just uses the signal processing to improve the image resolution, it has very important application value and broad prospects in astronomy, remote-sensing, medicine, business, video, military reconnaissance, security monitoring and other fields.This paper mainly focus on currently the most widely used POCS super-resolution image reconstruction algorithm, involved in image interpolation technology, multi-frame image motion estimation, reconstruction algorithm. Main tasks are as follows:(1) Studied the traditional image interpolation methods and adaptive image interpolation methods, through the experiment compared and analyzed the impact of image interpolation methods on the results of POCS super-resolution image reconstruction, and selected the appropriate image interpolation method for super-resolution reconstruction.(2) For the existence of the limitations of conventional motion estimation, based on the In-depth study of a variety of motion estimation, this paper designed hierarchical motion estimation algorithm in order to more effectively carry out motion estimation. And then the experiments on various motion estimation algorithms have been done and the experimental results have been analyzed.(3) In the standard POCS super-resolution reconstruction algorithm, there is the edge of oscillation and noise amplification phenomenon. To alleviate this problem, this paper used the improved POCS super-resolution reconstruction algorithm. Analyzed the number of iteration, the threshold size, projection operator and other factors that affect the result of POCS reconstruction algorithm, and the experiments have been done.