New distortion measurement methods for improving compression performance in motion-compensated hybrid video coding

Abstract

In the video coding standards of MPEG-x and H.26x, a motion-compensated prediction technique is used for enhancing the coding performance. This technique takes advantage of the correlation between consecutive frames in the time domain, which is relatively higher than that between adjacent blocks in the spatial domain. The conventional motion estimation and compensation algorithms adopted the block partitioning, fractional-pel-accurate motion estimation and multiple reference frames and so on. In the general reference encoders, these motion estimation and compensation algorithms are utilized for P- and B-frames, which are aligned as a fixed group of pictures (GOP). Even though these techniques had enhanced the coding performance significantly, these techniques are still not appropriate to the global- or local-illumination changes and scene changes. Therefore, we need new distortion measurement methods for motion estimation and scene change detection. The proposed distortion measurement method for motion estimation searches for a reference block that minimizes the motion-compensated residual signal when the DC-component is predicted. Also the proposed distortion measurement method for scene change detection calculates the entropy value in the difference frame and decides frame type for improving coding performance and suppressing the image quality fluctuations.