Research On Iterative Algorithm Of Image Restoration

As one of the most important media for receiving and transmitting information,images play a vital role in the daily life of modern people,aerospace,medical diagnosis,military public security,and other scientific fields.However,in practical applications,image degradation is caused by the imperfection of imaging system,recording equipment and transmission medium.Therefore,image restoration technology is particularly important,that is,the best approximation of the original image is restored according to the image degradation model and iterative algorithms.In this paper,the iterative regularization method for image restoration is analyzed and studied in depth.The main work is summarized as follows:(1)For the restoration of blurred images,the theoretical basis of image degradation,the research background and significance of image restoration and the current research status at home and abroad are described in detail.(2)Based on the mathematical model of image degradation,the ill-posedness of image restoration,the classification of degradation functions,and the classification of image degradation noise are analyzed.The basic theory of the regularization method is studied,and the process of the discrete degradation model of the medium rectangle formula is derived.(3)Based on Tikhonov and TSVD regularization,the RRGMRES and CGLS methods in the Krylov subspace method are studied.The selection strategies of three regularization parameters that affect the regularization effect are mainly discussed,which are Morozov’s deviation principle,generalized cross validation(GCV)criterion,and UPRE criterion.(4)Based on the Krylov subspace method,a hybrid regularized LSQR algorithm formed by Tikhonov regularization combined with the LSQR algorithm is proposed and applied to image restoration.Through numerical simulation,it is analyzed and compared with other Krylov subspace methods.The results show that the proposed method has the advantages of high SNR and small relative error.(5)Starting from the iterative regularization method,a regularized GMERR algorithm composed of TSVD regularization and GMERR method is proposed.The feasibility and effectiveness of the regularized GMERR algorithm can be obtained through numerical simulation and comparison results,and the quality of image restoration can be significantly improved in image restoration applications.