Optimization And Implementation Of AVS Mode Decision Algorithm

Video compression codec is the key technology of digital multimedia processing,and playing very important role in many fields, such as: digital TV, HDTV, videoconference and so on. AVS used in this paper is the second generation of videocoding standards of having the independent intellectual property rights in our country,and in2006it was approved by the national standards promulgated. AVS introduced aseries of codec technology, compared with the international mainstream video codingstandard such as H.264, MPEG-2, which balances the compression efficiency andcoding complexity and achieves the characteristic of high performance. At the sametime AVS solves the problem of H.264、MPEG-2standard which is encumbered bypatent and the problem of industrialization will cost too much. Therefore, researchingon AVS and implementing it have very important value and significance, and canbring about considerable economic benefit and social benefit.The paper aims at optimizing mode decision module of AVS codec standard onTMS320C6455hardware platform, which mainly involves looking for the fastalgorithm of this module, taking into full consideration and making full use of theresource of hardware and software and eventually speeding up the mode decisionmodule.This topic chose TMS320C6455as AVS video encoding hardware platformwhich is fixed-point DSP of high performance and is applied to many field such asvideo processing、 wireless communications and so on.TMS320C6455has thirdgeneration high performance VLIW architecture,9600MIPS operation ability andabundant peripheral interface, is the ideal platform of video codec.The main content of this paper is optimization and implementation of AVS modedecision module. Firstly, introduced the importance of mode decision module ofvideo codec technology and what has researched in many coding standards (AVS,H.264...);Secondly, analyzed the algorithm of this module from standard andreference code, proposed the module of the fast algorithm, and then optimized code、test the code. Thirdly, descript the principle of TMS320C6455hardware developmentplatform, available resources and code optimization methods and presented manyoptimization methods on the code and the platform, such as data types, compileroptions, inline function. The simulation results showed that the consumed time of the proposed algorithmin this paper was less20%-30%than the previous mode decision method while thesubjective visual quality and PSNR value did not decreased. Finally, the proposedmode decision algorithm was transported on the TMS320C6455and reduced thecoding time more than30%.