An adaptive edge-preserving artifacts removal filter for video post-processing

We propose an adaptive edge-preserving artifacts removal filter (EPARF) for post-processing video compressed using block-based motion compensation so as to improve visual quality. The proposed filter is based on half-quadratic regularization, and it can adaptively remove both blocking and ringing artifacts while preserving object edges. The proposed algorithm has been applied to video sequences compressed by the MSRA scalable video coding (SVC) codec with block-based motion-compensated temporal filtering (MCTF). Experimental results demonstrate that EPARF can significantly improve both subjective and objective quality.