Research On Entropy Coding Of AVS And Realization On DSP

With the emergence of communication networks, along with the rapid growth of multimedia service, the role played by multimedia communication has become increasingly important and essential in next generation communication system. Video coding is the fundamental and kernel tech in emerging multimedia/video communication. With the rapid development of digital multimedia technology, the complex fast and efficient coding algorithms have become the key technique for both video and image transmission.The main data compression methods are introduced firstly,the key technologies and characteristics of the most famous video coding standards, like MPEG series, H.26x series and AVS are analyzed. Based on the theory of video compression technology, the coding frame and principle of AVS has been researched , then AVS is compared with H.264 and MPEG-4; the AVS entropy coding is realized aiming at increasing the coding efficiency.Then a simple bit-rate estimation model is proposed, which makes entropy coding unnecessary in the rate-distortion optimizing process. This model can estimate approximate bit rate through some syntax elements directly, therefore, reducing entropy coding cost. Experiments show that the algorithm can reduce the coding complexity and save coding time with acceptable loss of video quality. It is convenient for hardware implement.Being implemented on the DaVinci DM6446 platform, in response to AVS 2D-VLC entropy coding features, some optimizing techniques, including C code optimizing, intrinsic functions, and linear assembling code, are applied to the entropy coding module, exponential Columbus code algorithm and write bit stream process, which makes the total cycles needed decreased by nearly 50%.At last, conclusion of this thesis and the discussion of future research directions are given.