Study On Image Restoration Algorithms Based On The Prior Information

Classical algorithms of image restoration include the inverse filter, wiener filter, constrained least squares filter and so on. In this thesis, blur image restoration methods based on partial prior information and constrained least squares algorithm based on orientation information measure are illuminated in detail.Blur image restoration based on the partial prior information has been developed in this paper , and gets PSF using a rule of correspondence between the partial prior information with the blurred image. First, a finite support image of prior information is got. Second, a finite support blur image matching to partial prior information is obtained with image matching algorithm based on correlation coefficient. Third, true blur image corresponding to prior information is acquired with iterative method of enumeration. Fourth, priori blur identification gets PSF by Wiener Filter method. Finally, entire image is estimated with wiener filter method.This paper puts forward constrained least squares algorithm based on the orientation information measure too. We create edge restriction using edge-preserving property of orientation information measure, edge-information is considered to a novel constrained condition, and parameter of noise and edge-information is estimated by generated cross-validation (GCV) method. Noise is smoothed and edge information is preserved in result of the image restoration, it is better adapted to human's visual characteristic.On the other hand, several ideas and future work are written in the end of the paper based on the conclusion of forenamed methods.