Temporal Scalable Video Transmission Using Multi-Reference Prediction Chain Coding

XS Lin,QH Dai
DOI: https://doi.org/10.1109/icme.2004.1394635
2004-01-01
Abstract:The work presents a novel multiple reference picture selection scheme, named multi-reference prediction chain, to encode video. Temporal scalability is obtained from the prediction scheme in which prediction chains are constructed for the pattern of prediction dependency between P-frames. The inter-chain independence also endows the video with error recovery ability. We dynamically manage the prediction chains or a group of frames using a rate-distortion optimization framework that adapts to the channel bandwidth so that the expected end-to-end video quality may be achieved under a given rate constraint. The expected error resilience capability may be achieved by the entire recovery of the error chains. In our algorithm, the video is pre-encoded to be transported partly with temporal scalability. Simulation results show that our algorithm is feasible in temporal scalability and has high performance in error resilience.
What problem does this paper attempt to address?