Study On Blind Restoration Algorithm For Blurred Image

In the imaging process,images generally need to go through many intermediate links,such as acquisition,storage and transmission.In addition,the quality of equipment,imaging environment and other factors influence the quality of acquired images,which is also called image degradation.The degraded image is relatively fuzzy,the noise point is prominent,the resolution is insufficient,and part of the regional information is lost,which leads to many obstacles in the subsequent image processing.Image restoration technology is an effective solution to the above problems.According to whether the Point Spread Function(PSF)in the corresponding degradation model is known or not,image restoration can be divided into two types: one is the known PSF non-blind restoration,and the other is the unknown PSF blind restoration.Because the PSF of the blind restoration and the clear image are not clear,both need to be estimated,so the blind restoration of blurred image is a kind of ill-posed inverse problem.Blind restoration of fuzzy images generally requires two steps.The first step is to solve the problem of blurred kernel prediction,and the second step is to recover and process the target image according to the predicted fuzzy kernel with the help of corresponding defuzzifica means.This paper focuses on the operation of fuzzy algorithm,blurred kernel prediction,correlation correction algorithm,etc.Aiming at the problem that the classical iterative blind deconvolution algorithm is more prone to defuzzification and slow convergence,this paper proposes a corresponding improvement method.Firstly,the bilateral image is used to preprocess the blurred image.Secondly,the information entropy algorithm is used to calculate the limited support domain of the image.The iterative replacement of the space and the frequency domain is performed in the support domain,which effectively solves the problem of slow convergence and defuzzification.The experimental results show that compared with the classical iterative blind deconvolution algorithm,the improved algorithm recovers the peak signal-to-noise ratio of the image,and the algorithm converges quickly and the recovery effect is better.The blind restoration problem of motion blurred images is mainly studied.Aiming at the blind restoration algorithm of motion blurred image,the fuzzy kernel estimation and fuzzy kernel correction are analyzed.The rolling guide filter and the operational gradient map are proposed,and the fuzzy kernel is modified repeatedly by iteratively.Then the LR algorithm and the modified fuzzy kernel are used to recover the image.Methods.Since the method of the present invention introduces a processing model for saturated pixels,the ringing effect caused by the region with relatively high luminance in the restoration result is greatly reduced,and the effect of the restoredimage is greatly improved.