Research On Error Resilient Transcoding Of Video Transmission In Wireless Network

With the technological development of computer science and wireless networks, video information processing and transmission in the wireless network are becoming more and more concerned. Wireless networks cannot provide reliable data transmission because of the high bit error rate, packet loss, limited bandwidth and multi-path fading. The most disadvantage of the compressed video stream is that it is sensitive to the error while error propagation is a common phenomenon. As a result, how to improve the video error resilience becomes an important research content, error resilient video transcoding is an effective method to solve this problem.This paper take the basic feature of the video stream and the characteristic of the wireless network into consideration, and take balance of the coding efficiency and error resilience. The dissertation makes intensive research on error resilient video transcoding in wireless environment, the main work and innovation points are as follows:First, the paper makes an introduction of the technology of the video transcoding and the error-resilient algorithm. This paper research the characteristic and model of the wireless channel, builds the simulation platform over 3G wireless network and compare some error resilient strategies on this platform.Second, an intra macroblock refresh algorithm based on motion vector is proposed in this paper. The algorithm firstly gets the motion vector information from the decoded video, then calculates the number of the refresh macroblock each frame by combining the wireless channel information from the feedback. The experiment results show that the algorithm can effectively improve the error resilience of the video stream without obviously bit rate increasing.In the last, the paper proposed an improved refresh algorithm based on rate-distortion optimization to make full use of the limited bit rate resources and adapt to the time varying wireless network. The new algorithm determines that a macroblock should be refreshed by rate-distortion cost, it can refresh macroblocks dynamically. The experiment results show that the algorithm can further improve the quality of the tanscoded video and reduce the output bit rate.