Based On The Dm642 Avs Real-time Encoding To Achieve Optimized

The AVS standard is the series of "information technology advanced Audio and Video Coding" standard abbreviation. The AVS standard includes four main technical standards: system, video, audio and digital copyright management, and support standard: consistency test. The fully Intellectual Property of second generation Source Coding-Decoding Standard - AVS is owned by China. Source coding technology mainly focuses on mass video and audio (original data and sources) compression technology, also known as digital video and audio coding technology. Obviously, as the premise of digital information transmission, storage and broadcast it is also the general foundation character standard of the digital video & audio industry. On 30th April, 2005, AVS standard video part officially passed the public show and became the national standard. AVS coding efficiency is more than 2-3 times of MPEG-2, and similar with AVC. Moreover the AVS technical plan is succinct, chip is less complex. It also achieves the highest level of second generation.AVS has obvious superiority in terms of coding efficiency and channel adaptive capacity, compared to those existing video compression standards. However, commonly used Platforms can not meet the needs of the high computational complexity. With its strong Parallel Processing ability and signal processing capacity. TMS320DM642 of TI Corporation is an ideal Platform to run AVS CODEC. Based on the thorough AVS standard and reference code, this dissertation Proposes techniques to port and optimize AVS encoder from PC to DM642 Platform. Some experimental results are presented to verify the effectiveness of the optimization.This dissertation will make an emphasis on research of key technologies and the implementation of AVS video encoder based on DM642 platform. First, we introduce the algorithms in AVS standard. Then by analyzing the tradeoff between performance and complexity of each module, we make some improvements and simplifications to the algorithms. The motion-estimation always occupies a very large percentage in the whole encoding procedure. An improved block matching algorithm and an improved inter mode select algorithm are proposed to improve the code efficiency. This dissertation designs a new diamond search (DS) algorithm. Finally, we discussed the method that AVS encoder based on PC is transferred to DM642 platform, and then the application optimization in DM642 embedded system was made by the following sides: application architecture, memory system, Cache optimization and EDMA, key functions line assemble. After those steps, not only the encoding performance is improved, but also many skills of software development about DM642 were summarized.Experiments validate that the encoder can realize real-time video coding at the speed of 30 CIF pictures per second. Moreover it possesses good visual quality and high compression ratio.