Non-blind Deblurring Of Color Images Based On Nonlinear Diffusion Equation And Curvature Term

Image deblurring is an important research direction of digital image processing.The blurring of images can cause great inconvenience for identifying and analyzing the content of image information.In order to get rid of the blur,many scholars make a lot of contributions.In order to solve the ill posed inverse problem,a variety of blurry image restoration methods are proposed.Whereas one of the important methods is based on a variational optimization problem where the desired solution minimizes a criterion composed of fidelity and penalty terms,and the the Total Variation(TV)regularization method is representative of this type of method.The Total Variation regularization item can control the smoothness of the restored image and can preserve the edge detail information of the image better.However,the method can not maintain the detailed information after processing the blurred image,and have some problems such as visual disconnection occur.In order to improve the effect of deblurring,this paper replaces the rule items in the TV regularization model with nonlinear diffusion rule items and curvature items,and uses the characteristics of nonlinear diffusion and the characteristics of curvature items to highlight the details to repair the image information,and process the image by extending the single channel to the multi-channel.In order to improve the computational efficiency of the model,a fast iterative algorithm-Split-Bregman algorithm is used to solve the non-blind deconvolution image.Through the subjective visual effects and objective experimental numerical analysis of real images,it is shown that the model in this paper is robust to fuzzy noise for a single color blurred image,can improve the image restoration speed,effectively avoids "ringing effect" and "ladder effect",and can obtain better restored images.