Our proposed system contains two stages named as detection of defects and removal of degraded pixels. Adaptive histogram shape based thresholding is applied for creating defect map(finding the degraded pixels).The quality of degraded pixels will be enhanced by constructing image prior structure over extended image neighborhoods. We give new solution for the challenging task such as modeling image priors, due to the high dimensionality of images, their non-Gaussian statistics. We have to model correlations in image structure over extended image neighborhoods. We quantify our performance by computing the quality measures such as PSNR and MSE.