Adaptive edge-preserving smoothing and detail enhancement for video preprocessing of H.263

This paper proposes an adaptive edge-preserving smoothing and detail enhancement for video preprocessing of H.263. The proposed method consists of two steps: edge preserving smoothing using a motion-adaptive bilateral filter and detail enhancement using an adaptive high-boost filter. In the first step, the motion-adaptive bilateral filter preserves edges in region of interest (ROI) and smoothes noise in non-ROI, so that it can make video coder be more efficient under the restricted bitrate. Classification of ROI/non-ROI is done by simple thresholding of the absolute difference image between two adjacent frames. In the second step, an adaptive high-boost filter is used to enhance spatial edges of the decoded image. The bilateral filter and high-boost filter require user parameters according to the amount of motion and edgeness, respectively, and perform proper filtering with adaptive parameters. Experimental results show the effectiveness of the proposed method in terms of the visual quality and the peak signal-to-noise ratio.