A Research Of Motion-blurred Images Restoration Technology

Image restoration is an important topic in image processing. Images are degraded for many reasons, among these reasons the relative motion between the target image and a camera can’t be ignored. So the restoration of motion blur images is an important branch. Motion blur is very general in the process of acquiring an image. The pictures acquired from the planes and photographic spacecrafts, images with high speed all have the problem. So the restoration of motion blur can be widely used to Space project, military secrets, automation control, medical imaging, traffic monitoring and public safety. So it has important and practical significance. Yet there are many relative mature recovery methods been put forward by the researchers, but to the different types of motion blur, nobody has put forward a general recovery method, motion blur restoration is still an important and difficult problem to be solved.This paper is based on the analysis and comparison of the previous study of motion blur restoration, and focuses on the research of the uniform linear motion blur. It realizes the design of two types of motion blur parameters identification alternately; they are based on radon transformation and alpha channel separately. To the former algorithm, this paper makes a simple improvement to improve the accuracy of motion parameter estimation. To the ringing effects that appear in the image restoration process, this paper completes the elimination of them. Finally, applying the motion parameters based on the alpha channel identification algorithm to the non-uniform linear motion blur.The work In this paper is as follows:(1) Discussing and analyzing of the purpose and significance of the thesis, summing up the development and research of the motion blur image restoration technology, over viewing of the main work of this paper.(2) Researching on the interference and elimination the image has suffered from outside, compared to several classical filtering algorithm and the popular filter denoising algorithm, and summarizes the advantages and disadvantages of these algorithms and their application of occasions; Analyzing several commonly used image restoration algorithm, selecting better performed recovery algorithm for the following of the recovery work.(3)Focus on the research of the uniform linear motion blur degradation model. And prove the validation of the model through reduction. Completing the research and improvement design based on radon transform algorithm The simulation results and the comparison with the references verify the effectiveness and superiority of the proposed algorithm.(4) Summarizes the existing common solution to the local motion blur method, and finally select the restoration algorithm based on the alpha channel. Utilize the technology of digital matting, motion blur constraints, the RANSAC algorithm and the EM algorithm to estimate motion parameters comprehensively. Reduce the amount of computation, save processing time through the user interaction.