Research On Distortion Estimation Model For Witsenhausen-wyner Video Coding

In recent years, with the development of science and technology, the videoencoding technology also has an explosion of development. The existing videocoding standards, H.26x/AVC, can be well done in real-time transmission andstorage of large-scale video, which makes long-distance video transmission, videotelephone, video conference, interactive multimedia applications becometrue. However, in the wireless network, the compressed bit stream duringtransmission cannot be avoided the impact of the external environment. Firstly, thewireless network environment is complex and changeable, the network topology isconstantly changing, resulting in changes of the actual available bandwidth;Secondly, the surrounding environment in the wireless network is vulnerablebecause of the impact of weather, geological activity, the interference ofnoise. These all lead to the decoder receives the part of stream which cannot berestored to get the desired effect.To avoid this problem, Mei Guo from Harbin Institute of Technology hasproposed a new video coding framework: Wytsenhausen-Wyner Video Coding(WWVC) Framework. And the framework effectively improves the video decodingeffect in the wireless network conditions. The new video coding framework, WWVC,has combined the advantages of the traditional video coding and distributed videocoding, which ensure real-time video streaming stable and smooth in wirelessnetworks. At the same time, the compression effect is only slightly lower thanH.264/AVC.Although WWVC has got very good results, there are still some problems. Firstand foremost, the mode decision of WWVC ignores the important factor (packet lossrate) to video coding framework in the transmission network.To solve the above problems, the paper first introduced the packet loss rate inthe mode decision, changing the distortion estimation model from the traditionaldistortion estimation model into the end-to-end distortion estimation model. ROPEand REED are two typical end-to-end distortion estimation models. In our proposedmodel, we have made a good balance between the complexity and compressioneffect. The study proves the proposed model has a good influence on video codingperformance and increase the robustness of the framework.