Research On The Methods Of Blind Image Restoration

In the process of capturing images, they are degraded by aberration, diffraction and nonlinear aberrance of imaging system, nonlinearity of film sensitization, relative motion during imaging, atmospheric turbulence effect, stochastic noise and so on. In order to eliminate the effect of degradation and restore the features of image, we need to use image restoration technology. But in practice, the priori knowledge is always limited, we need to extract information from the degraded image. The method which only usesthe data of degraded image to restore image is the main content of the paper——BlindImage Restoration.This paper introduces the application background and basic methods of blind image restoration briefly, analyzes the degradation model of image and the ill-posed problem during restoration in detail.Based on the theory of constrained least square, this paper presents a blind imagerestoration method——space domain conjugate gradient algorithm. We establish a newcost function to incorporate the information about degraded image and point spread function (PSF), then use the conjugate gradient algorithm to search optimum solution. The deblurred image and the PSF are well simultaneously estimated with the algorithm.The restored image obtained by using the space domain conjugate gradient algorithm is too smooth and loses some information about the edge detail. In order to solve the problem, we research into the regularization parameter and present a new method of adaptive selection of the regularization parameter (ASRP). This method dispenses with the priori knowledge of noise variance. It can obtain the regularization parameter adaptively according to the edge information of the image, make sure the restored image is the best result in the total field, and keep the edge detail of the restored image well.But when the degraded image has both noise and blur, the restored image using ASRP method is not perfect. In order to restrict the noise effectively, the paper adopts regularization image restoration method based on the anisotropic diffusion theory (RBAD), achieves noise restriction and detail preservation simultaneously. The restoration result is improved much better. The author not only analyzes the theory, but also uses Matlab software to do simulated experiments about the proposed methods. The experiment results show the methods are all correct and feasible.