Regularization Strategies Of The PCG Algorithm In Image Restoration

Image restoration is one of the most important problems in the image processing.In order to solve deconvolution problems with noisy data,the adoption of PCG Algorithm had became an obvious trend.In accordance with the large scale and badly Ill-condition of image restoration,this thesis presents a class of regularized PCG algorithm (abbreviated as RPCG) which combines PCG algorithm and some regularization strategies.These strategies include the construction of new preconditioner,the adoption of optimum stopping rule and the extrapolation techniques of RPCG-solution with respect to regularization parameter.Because of the good suitability of the above techniques to parallel computation,the related paralled Algorithm is also given.To show the validity of the algorithms,an applied software is developed under the enviroment of Matlab,and numerical tests are also made.Theoretical analysis and numerical simulation indicate that the new algorithm possesses the advantages of high-speed,good stability arid the inhibility function against the noise in blurred image.