Multiple Reference Motion Compensation: A Tutorial Introduction and Survey

Abstract

Motion compensation exploits temporal correlation in a video sequence to yield high compression efficiency. Multiple reference frame motion compensation is an extension of motion compensation that exploits temporal correlation over a longer time scale. Devised mainly for increasing compression efficiency, it exhibits useful properties such as enhanced error resilience and error concealment. In this survey, we explore different aspects of multiple reference frame motion compensation, including multihypothesis prediction, global motion prediction, improved error resilience and concealment for multiple references, and algorithms for fast motion estimation in the context of multiple reference frame video encoders.

A survey on video compression fast block matching algorithms

“…Motion vectors are encoded by lossless compression, while RPE is encoded by lossy compression to get high compression ratio [9], [83], [13], [110], [92], [4 ].…”

“…The MVs are encoded using entropy coding and RPE between the current frame and the MCP is encoded using transform coding, quantisation and entropy coding. At the decoder, the received MVs will be utilised to form an MCP from the reconstructed reference frame, and then the current frame will be reconstructed by adding the reconstructed RPE to the MCP [9], [83], [13], [110], [92], [4].…”

Enhanced Computation Time for Fast Block Matching Algorithm

“…The first output of this process will be the difference between the current frame and the MCP, which is called the Residual Prediction Error (RPE) (or Displaced Frame Difference (DFD)); the second output will be the motion vectors. The MVs are encoded using entropy coding and RPE between the current frame and the MCP is encoded using transform coding, quantisation and entropy coding, as shown in Figure 1.2 [Sullivan et al, 2004;Leontaris et al, 2009;Richardson, 2010;Sayood, 2006;Marpe et al, 2006;Al-Mualla et al, 2002]. At the decoder, the received MVs will be utilised to form an MCP from the reconstructed reference frame, and then the current frame will be reconstructed by adding the reconstructed RPE to the MCP [Bhattacharyya and Deprettere, 2010].…”

“…Motion vectors are encoded by lossless compression, while RPE is encoded by lossy compression to get high compression ratio [Sullivan et al, 2004;Leontaris et al, 2009;Richardson, 2010;Sayood, 2006;Marpe et al, 2006;Al-Mualla et al, 2002]. This thesis focuses on this stage and the details will be introduced in sections 3.2 and 3.3…”

Motion compensation for video codec based on disparity estimation

“…Where f n (i, j) is the pixel in frame n at position i, j and f n−1 (i, j) is the pixel at position i, j in frame n-1 and ˆ f n (i, j) is the prediction value of the pixel at position i, j in frame n. [4] The effects of the algorithm are shown in Figure 2 (a), (b). For still cameras or local motion the algorithm get competitive results since the difference between two neighbor frames is very small and the compression will lead to very small size output video files.…”