Motion Blur Identification Using Image derivative

Abstract

Motion blur is one of the most common blurs that degrades images. To restore such images, a good estimation of motion blur parameters is necessary. In this paper, we have proposed a precise algorithm to estimate blur extent in linear and non linear motion blur for gray scale images. To estimate blur extent, the images were rotated such that motion blur was in horizontal direction. The proposed algorithm is based on analysis of image derivative along motion direction and processing the areas of the blurred image around an edge. The experimental results were great in the case of noiseless and noisy images.