Resilient Video Coding via Improved Motion Compensated Prediction

Abstract

Motion compensated prediction (MCP) is at the heart of modern video compression standard because of its ability to remove temporal redundancies. However, MCP is responsible for temporal error propagation, which can result in severe quality degradation in lossy environments. In this chapter, the authors present two innovative methods of improving MCP to be more resilient to packet losses. In the first method, the motion trajectory is used to develop a novel distortion weighting technique, and the second method exploits the presence of Intra macroblocks in previously coded frames to develop increase robustness.