An Edge-Based Adaptive Post-Processing Technique for Block-Based Discrete Cosine Transform Coded Images

In this paper, an edge-based adaptive post-processing technique is proposed, with the aim to remove the blocking artifacts as well as ringing artifacts especially in the situation where edges in the coded image degrade seriously. The proposed algorithm takes two main steps: deblocking step and deringing step. In the deblocking step the blocks are classified as low frequency or high frequency blocks first, then different methods will be adopted to remove the block artifacts according to the flatness along the block boundary. In the deringing step, the image is divided into three parts: main edge vicinity area, detail edge area and flat area according to the result of edge detection. Then different filters will be used to different areas. In this way, both the artifacts beside the strong edge and detail edge will be removed and it will not introduce new blur. Simulation results show that the subjective perceived quality as well as objective evaluation is improved.