Block-edge reduction in MPEG-1 coded images using statistical inference

A method is proposed in this paper to filter the block-edge artifact which is a common problem in block-based coding mechanisms used in various digital image and video coding standards. In this method, an algorithm for post-filtering is proposed, based on local statistical characteristics of a video image. This algorithm makes the filtering process adaptive when filtering the block-edge artifacts at different quantisation levels using a linear filter. In the filtering process, the proposed method first determines the blocks causing high block-edge artifacts and then a linear filter is applied along the edges to reduce the block-edge artifacts.