率失真优化时域采样的多描述视频编码 2.docVIP

 Multiple Description Video Coding Using Rate-Distortion Optimized Temporal Sampling# 5 10 15 20 25 30 35 40 45 Huihui Bai, Lili Wan (Institute of Information Science, Beijing Jiaotong University) Abstract: In this paper, an effective multiple description video codec is designed based on pre- and post-processing of video sequences without modifying the actual coding process itself, which makes 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 a good estimation of lost frames in post-processing after side decoding. In particular, given a central distortion, rate-distortion optimization with La-grangian formulation is applied in pre-processing to achieve a good tradeoff between bit rate and side distortion. Furthermore, for simplicity of the Lagrangian multiplier selection, a new optimization criterion is proposed based on the correlation of intra-/inter-description. The experimental results exhibit better performance of the proposed scheme than other tested schemes. Key words: Video coding, multiple description coding, Temporal sampling 0 Introduction For the compressed video sequences, due to motion prediction and compensation, any random bit errors or packet losses will cause substantial quality degradation. As a result, compressed video transmission over unreliable networks is a challenging problem. Multiple description coding (MDC) is an attractive framework to solve this problem. It can efficiently combat information loss without any retransmission thus satisfying the demand of real time services and relieving the network congestion [1]. MDC encodes the video into several bit streams (descriptions) carrying different information which can be transmitted over channels. The descriptions can be individually decoded to reconstruct video of a coarse fidelity which is measured by side distort