Multiple description video coding using correlation optimized temporal sampling

Huihui Bai, Mengmeng Zhang, Anhong Wang, Yao Zhao

Abstract: We propose a novel multiple description video coding scheme based on pre- and post-processing of video sequences without modifying the actual coding process itself, thus making it compatible with the current standard coding. In pre-processing, adaptive modes of temporal sampling are employed to regulate the motion change between frames before performing odd/even frame splitting, which facilitates good estimation of lost frames in post-processing after side decoding. Significantly, given a central distortion, rate-distortion optimization with Lagrangian formulation is applied in pre-processing to achieve a good tradeoff between bit rate and side distortion. Furthermore, to simplify Lagrangian optimization, a new optimization criterion is proposed based on the correlation of intra- and inter-descriptions. In the experiments, the proposed scheme achieves better performance compared with other tested multiple description coding (MDC) schemes in both an on-off MDC environment and lossy packet networks.