Research On Real-Time Video Coding And Transmission Over Wireless Channels

With the rapid growth of multimedia content in today's communication links, there is an increasing demand for video coding schemes and underlying transport that can provide acceptable quality of service to the end users. Due to the errors nature of channels and the error propagation property of compressed video, suitable source and channel coding schemes are required to handle such conditions.This dissertation deals with the issues of video coding and transmission over unreliable channels. Based on the video coding standards, such as MPEG-4 and H.26L, and networks protocols, we presented some schemes on error resilient video coding and transmission techniques. The main work includes:1. Two improvements have been made for encoder error resilience based on the analysis of the code stream structure of MPEG-4. One is achieved by not using the AC prediction for DCT while Intra coding, the other by encoding the all four modes of MB of B frame. These two improvements have been proved to enhance the error resilience without decreasing the encode efficiency by experiments.2. An error detecting method which can prevent the decoding error affecting the next video packet head is used, and a restoration measure for B frame based on the MV prediction which can improve the visual quality has been studied.3. A novel marking algorithm for IP DiffServ based on the transmission condition and the different importance of video streams for decoder has been introduced. In the border router, we design a new DiffServ scheduler(BSWCS: Bit-based Weight time Slot Compensate Scheduler),which dynamically adjust the weighted quantum of service. Combined with the error concealment technology in the video decoder, in case of severe congestion which cause losing a lot of IP packets, these algorithms have been proved effective, hence the delay, jitter and the perceived picture quality all meet requirement for real-time video transmission.