Non-Iterative Adaptive Post-processing for Ringing Artifact Suppression in Compressed Images

摘要

This paper presents a simple post-processing algorithm for suppressing ringing artifacts that are present in compressed images. The behaviors of ringing artifacts are first studied. It is found that the magnitude and pervasiveness of ringing is dependent on the local spatial activity and the compression rate of hte image. Ringing is more severe and visually annoying around strong edges and in highly compressed images. Based on these observations, we propose an adaptive post-processing technique using a selective weighted mean filter (SWMF), which performs different levels of smoothing over different region sizes. Simulations showed that the proposed algorithm is effective, and does not result in blurring of edges and texture. It is also fast as it employs only non-iterative processing.