Research On Motion Blurred Image Restoration In Atomization Scene

In recent years, there have two aspects of quality degradation problems in outdoor image shooting, on the one hand is the weather factors that the large-scale fog appearing all over the country, even the haze days, and make the image have a phenomenon of atomization. On the other hand the relative motion between filming equipment and scenery will appear, and make the image motion blur caused by mass reduction. These two problems not only affect the image visual effect, and will cause the image loss important part of the details; the analysis of the image will become very difficult too. In this paper, we will both make the image restoration and the atomization of motion blurred, and to achieve the effect that restore both the atomization and fuzzy. In this paper, the research mainly revolves around the following three parts.This paper will eliminate the image restoration of atomization based on dark channel prior algorithm in the process of image to atomization. The algorithm can estimate the atmospheric light and image transmission rate, but the estimate to the transmittance is rough, there has block effect in rough transmittance image, edge or image is not smooth too, the image will produce the phenomenon such as halo and color distortion after restore. In this paper, we optimized the rough transmissivity based on CDD model, the optimized transmittance increases in edge smoothness, plaque effect is decrease in the number and size. And then combine the transmission rate that optimized and the solution of the atmospheric optical and the fog image degradation model to make the regression atomization. In this paper, we compare the results between transmittance image optimized by CDD model and the transmittance image deal with Orientation filter, we can see that the restoration of CDD model can doing well with the edge of the transmittance image, and the CDD model is better than other methods and to improve the fog effect well from simulation experiments.We need to estimate the point spread function in motion blurred image restoration algorithm in the first, it contains two important parameters:motion blur Angle and length of motion blur. This paper mainly introduces the estimation problem of point spread function, and put forward the optimization method of estimation method of PSF function:First doing the edge detection processing to the motion blur image, and extract the edge of the blur image, and then making the block processing to the image, keep the testing image blocks that have more edge information, then find the original blocks that corresponding to the original degraded image, and estimating the blurred angle and the blurred length of the original blocks, and we will got the optimized blurred angle and the blurred length from the mean values of the closest two angles and two length value. We use the classic wiener filtering method to recover the image after the point spread function be calculated. It can be seen from the simulation that the optimized method improves the accuracy in the estimate of PSF and making the recovering effect increased.Finally, we restore the image taking outdoor that contains both atomization phenomenon and the motion blur mass reduction, we deal with the atomized restoration of degraded image first, and then recover the motion blur based on atomization recovery, verify the practicability and feasibility of the algorithm in the image restoration results. The simulation results shows that the algorithm of atomization restoration and motion blur restoration is independent, the image motion blur has no effect to the atomization restoration, at the same time, the image motion blur have no interference to the estimation result of the blurred angle and the blurred length after the atomization restoration.