Single parameter post-processing method for image deblurring

Abstract

Numerous algorithms exist for the problem of deconvolution of blurred images. But due to ill-posed nature of deconvolution, many images still remain blurry after deblurring. An edge sharpening algorithm is proposed in the paper to further improve the quality of blurry images in edge areas. The method is based on pixel grid warping, its main idea is to move pixels in the direction of the nearest image edges. Warping allows to make edges sharper while keeping textured areas almost intact. Experimental analysis for different optical blur models is performed to optimize the parameters of the proposed method and to show its effectiveness.