Research On Blind Image Restoration Method Based On Multiple Prior Constraints

Blind image deblurring is a classic problem in the field of image processing and computer vision.Its goal is to recover hidden images hidden in blurred images.In the process of blind image restoration,we lack two key information of blur kernel and original image,so this is an underdetermined problem,that is,there are infinite solutions.In order to solve the underdetermined problem and obtain the numerical solution of the blur kernel,additional constraints and prior knowledge about the blur kernel and image are needed.Therefore,this paper proposes an effective blind image deblurring method based on multiple priors.The basis of our method is the fact that the restored image tends to be clear.Based on the current blind image restoration method,the restoration effects of some specific scene images,such as low-illumination images and background complex images,are not ideal,and there is some blurring(image outline and detail representation are not clear).This paper aims to solve these problems and proposes to combine multiple priors,including dark channel prior,intensity image prior and gradient image prior knowledge,and weigh them to provide more contours and details during the restoration process.A priori information,which is very helpful for blur kernel estimation.In this paper,we assign three kinds of priors to get a total prior knowledge,and put this prior knowledge into the framework of Maximum a posteriori estimation.The estimated blur kernel is obtained through continuous iteration,and the original image is restored by using the non-blind image restoration method.In the various scenarios in the generalization of the natural environment,our results have made great progress compared to the more advanced methods,and the outlines and details of the results have been improved.