Title :
A novel source-channel constant distortion model and its application in error resilient frame-level bit allocation
Author :
Yang, Hua ; Lu, Ligang
Author_Institution :
California Univ., Santa Barbara, CA, USA
Abstract :
Compared with either heuristic or RD optimization based approaches, frame-level bit allocation strategies based on constant quality models are inadequately studied. They are, however, inherently more capable of achieving consistent video quality over the whole sequence for real-time single-pass rate control tasks. Moreover, no existing model is error-resilient. In this paper, we propose a novel error resilient source-channel constant distortion (SCCD) model, whose uniquely selected frame complexity measure effectively accomodates the joint impact from both the nonstationary source video signal and the lossy channel. Compared with its existent counterparts, our model has a higher modelling accuracy, and can be generally applied in frame-level bit allocation to enhance error robustness of rate control. As an example, an error resilient frame-level bit allocation solution is proposed based on the model. Simulation results demonstrate that for lossy video transmission, the proposed method can achieve more consistent and higher overall visual qualities than that of the other nonerror-resilient schemes, which proves the remarkable potential of the new model.
Keywords :
bandwidth allocation; combined source-channel coding; error correction codes; image sequences; real-time systems; variable rate codes; video coding; SCCD model; constant quality models; error resilient bit allocation; error robustness; frame complexity measure; frame-level bit allocation; lossy channel; nonstationary source video signal; rate control; real-time single-pass rate control; source-channel constant distortion model; video sequence; Bit rate; Counting circuits; Decoding; Distortion measurement; Error correction; Loss measurement; MPEG 4 Standard; Propagation losses; Robust control; Video sequences;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 2004. Proceedings. (ICASSP '04). IEEE International Conference on
Print_ISBN :
0-7803-8484-9
DOI :
10.1109/ICASSP.2004.1326535