An improved method to estimate the motion blurred direction of motion blurred images

In the process of the image's recording and transmission, because of many factors, it could easily generate relative motion between the camera and objects and form a motion blurred image. So, how to recover motion blurred images becomes an important issue in the digital image processing. One of the typical approaches to estimate the motion blurred direction of the motion blurred images is based on the high-pass filtering. This paper is concerned with the drawback of this method that the speed of the estimation is too slow. The improved method increases the speed of estimating the motion blurred direction by automatically adjusting the parameter of step size and dividing the image into several parts. Three experiments are conducted to study the proposed method, and we analyze the results from three aspects: accuracy, speed and physical memory used by the MATLAB program for estimation. The experimental results indicate that our approach can effectively increase the speed of the estimation of the motion blurred direction with high precision, and reduces the physical memory used by the MATLAB program for estimation. Also, the proposed method has a good anti-noise performance. © 2012 Praise Worthy Prize S.r.l.