Scalable and Robust Video Compression

摘要

The research tasks are: Task 1, Scalable Coding; Task 2, Alternatives to variable length coding (VLC); and Task3, Packetizing for Error Concealment. Initial work on Task 1 built on research from a prior DARPA contract, and resulted in a scalable video coder with the unique property of scalable transmission at six resolution/frame-rate combinations. Task 2 work combined fixed and variable length coding (VLC) into a joint coding approach with distinct robustness advantages, while giving up only a slight amount of performance versus straight VLC coding in the noise-free case. Task 3 Packetzing for Error Concealment explored the new technique dispersive packetization and combined it with error concealment in a state-of-the-art subband/wavelet image coder. An ns-2 network simulation revealed substantial robustness to packet loss error, with useable images recovered under packet loss rates at high as 30%, thus implementing the graceful degradation property.