Font Size: a A A

Videophone Low Bit Rate Video Coding Algorithm

Posted on:2002-10-27Degree:MasterType:Thesis
Country:ChinaCandidate:Y LingFull Text:PDF
GTID:2208360032456692Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Recently the conventional communication methods such as telegraph and telephone can抰 meet the increasing need of people. Along with the development of multimedia, image processing and computer network technologies, most of the largest companies are marching into the market of low bitrate video products. Therefore the research of very low bitrate image coding has great significance.The author takes part in the research of the codec of ISDN videophone in Shanghai Bell Company and discusses the image coding algorithms in detail.To implement credible transmit of the videophone sequence image, the key problem is how to design an efficient video compression algorithm. Our research is completed on the H.263 framework. In this paper, the theory of image coding and the recommendation H.263 are introduced. Then the author describes several methods, which can improve the performance of the baseline H.263 encoder. All of these improvements have been implemented.DCT-based block image coding has always suffered due to block artifacts in the compressed image caused by individually processing blocks. In order to solve them, a new motion estimation algorithm based on macroblock partition is proposed in this paper, and a new rate control policy is adopted. Simulations show that the proposed algorithms can improve system coding efficiency and image quality. At the same time the author present a new non-linear edge-preserving filter, which yielded better results compared to the conventional filters.This paper also present a method that reduces the computational requirements of the encoder, while still upholding picture fidelity and remaining compatible with the H.263 bitstream standard. Often a substantial number of inter-macroblocks in the encoder are reduced to all-zero values after quantization. We introduced a method of predicting when those macrobolcks will quantize to zeros, which in turn allows us to eliminate the computation that would normally be required for those macroblocks.Lingyun(Computer Application)Directed by Professor Wang Linquan...
Keywords/Search Tags:image coding, low bitrate, block artifact, motion estimation, AZQ
PDF Full Text Request
Related items