Study And Optimization Of AVS-M Software Decoder Based On ARM9 Platform

Video compression, though a research focus brought forth by the fast development of multimedia technology and digital communication technology, has posed a higher demand on video coding due to the limitation of the bandwidth of communication and terminal processor.AVS(Audio Video coding Standard), the second generation signal source coding standard established by Audio and Video Coding Standard Workgroup, will be the solution to the lack of independent property right in Audio and Video compression domain in China. As the seventh part of AVS, AVS-M is the very mobile video coding standard according uniquely with the characteristics of wireless net.On the principles of AVS-M, reference code wm2.5a is first transplanted to ARM architecture, then c code non-distorted optimization on wm2.5a proceeded which explores, in two aspects of code & data size and decoding speed, the following kinds of methods as: redundancy elimination, memory movement reduction, structure adjustment, arithmetic optimization etc.. When different optimization methods were optional on one problem, the appropriate one was adopted according to the optimization result. Furthermore, the conformance test on the optimized decoder and the design on LCD output programming have ultimately perfected the decoder.Compared with reference code, the optimization result reduced code size and data size by 40.70% and 79.25%, and improved decoding speed by 5-8 times.When video test sequences (QVGA, above 384kbps) were decoded, the speed was 18～26fps on HHARM2440 (384MHz) platform, which basically answered the embedded system application requirement.