Research On Motion Blurred Image Restoration Algorithm

With the deepening of the application of image processing in various fields,the requirements for image processing are also constantly improving.In practice,due to the influence of many different factors such as relative motion,defocus and noise,the acquired images will be polluted to a certain extent and blurred images will be formed.Therefore,the processing of blurred images has become an important branch of image processing.Taking motion blurred image as the research object,firstly,the performance of two classical image restoration algorithms,inverse filtering restoration and Wiener filtering restoration,are analyzed.Based on the idea of distributed computing,a combined Wiener filter restoration method is proposed,that is,to eliminate some noise in the spatial domain first,and then to make Wiener filter restoration processing.Simulation analysis shows that the combined Wiener filter restoration method has better performance than the simple Wiener filter algorithm.However,the traditional algorithm has limited improvement space and cannot satisfy people's pursuit of higher quality images.In this paper,an intelligent image restoration algorithm based on sparse representation denoising based on region division is proposed.Firstly,the algorithm divides the whole image into non-smooth regions and smooth regions by using the Primal sketch sparse representation model and the statistical characteristics of image blocks.For non-smooth regions,the image denoising method based on K-SVD dictionary is used to denoise,and for smooth regions,the mean filtering method is used to denoise.Due to the idea that different regions use different algorithms,the algorithm retains more detailed information of the image.Under the same experimental environment,the PSNR value of the combined Wiener filter algorithm is improved by at least 1.4dB compared with the Wiener filter algorithm,while the PSNR value of the sparse representation denoising intelligent algorithm based on region division is also improved by at least 1.4dB compared with the Wiener filter algorithm.The simulation results show that both the combined Wiener filtering algorithm and the sparse ambiguity of degraded images to a certain extent and retain their effective detail information.The research results of this paper have certain reference value for the realization of motion blurred image restoration.