Motion Blurred Image Restoration based on Super-resolution Method

Motion blur is a typical and common degradation in surveillance system. The problem of motion estimation based on super resolution reconstruction of multiple images is addressed in this paper. This paper presents a motion projection onto convex set (POCS) algorithm to restore an image from multiple blurred images. The inter and intra pixel shifts are used to compute a joined point spread function (PSF), which determines the project model of the pixels in the moving window in each iteration. Experiments confirm the proposed algorithm outperforms single-image restored algorithms, especially for space-variant motion blurred images.