Research On Multiple Description Video Coding

Recently, the problem of packets loss and errors in video communication over erasure channels becomes an important topic. As a result, multiple description coding is proposed as an efficient scheme to tackle this problem. Multiple description coding is able to improve the robustness. This paper researches on several methods of multiple description video coding which are widely used in practical application. In combination with the latest video coding standard of H.264 and the specific network, previous schemes of multiple description video coding have been improved in this paper. The main contribution of this paper is as following:The unbalanced multiple description coding could give better control than balanced multiple description coding on the amount of added redundancy. In order to improve the coding efficiency, an unbalance multiple description video coding scheme which is based on SP-frame is proposed. SP frame which is a frame coding method, can reconstruct the same image sequences using different reference frames.With very small amount of redundancy, this scheme could limit error propagation in time domain during decoding processing by inserting SP frames in the descriptions. Furthermore, we propose an unbalanced multiple description video coding scheme based on region of interest segmentation. The regions of interest are more attractive than other regions when users watch image or video sequences. Firstly, we use an automatic modeling approach to detect attention regions in each frame and divide the frame into two parts. Then, the two parts are encoded into the high resolution description using modified video coding standard of H.264, and the low resolution description is produced by sampling the image sequences in the time and spatial domain. The simulation shows that the proposed method could effectively protect the data in the region of interest and gain better video quality.Cross-layer optimization method for multiple description coding over specific network of ad hoc is proposed. The ad hoc network is suitable for transmitting the multiple description data because it can provide multiple paths. We follow a cross-layer approach and formulate an optimal routing problem that minimizes the application layer video distortion. Objective function which is used to select three paths in the ad hoc network is presented. Furthermore, intelligent algorithm is used to work out objective function.