Implementation And Optimization Of AVS Decoder On OMAP3530

AVS(Digital audio video coding standard) is the second generation source coding standard with independent intellectual property rights in China, the video standard of which became the national standard in2006,March. Because of its high compression ratio, low system complexity, high image quality, low cost of patent, it has wide market application prospect. But at present, hardware and software for the AVS decoder are not mature, there is a certain distance away from industrialization promotion. Therefore, through the research and study of AVS video codec standard, a new generation of digital multimedia processor OMAP3530provided by TI company is used to design and realize an AVS video decoding system based on C64x+DSP, and the requirements of real-time decoding and playing of D1resolution is achieved by the late optimization. The main researc contents and results are as follows:First, this paper studied the AVS video decoding standard, analyzed and studied the main structure and process of decoder and main technique of AVS video decoder, mainly discussed the entropy decoding, inverse quantization, inverse transform, intra,inter frame prediction and interpolation module.Secondly, the AVS decoding system design and code implementation were realized according to the DSP development process, which combined with the overall framework of the system. And the image header, straps decoding, macro block decoding and image reconstruction of core module were gradually realized through the analysis of the key modules. Finally, the configuration of integrated development environment, emulator and target board debug, software and hardware simulation were all implemented.After the AVS decoder achieved success on OMAP3530platform, the post optimization was conducted, which included compiler directive, memory structure, C language and manual assembly language four aspects. Besides, this paper proposed a optimization scheme of entropy decoding module code table using the manual assembly, which aimed at the characteristics of VLC entropy decoding. It was an optimization of the hardware structure on DSP, which included the optimization of AVS index Kolumb code algorithm. The optimization was implemented by an inline function. Eventually, the running period of the module got greatly reduced.At last, we conducted performance test of the AVS decoding system. We tested and simulated the Dl format code stream on OMAP3530platform. After analyzing the data, a conclusion can be drawn:optimized decoding speed got greatly improved compared to original decoding speed under the precondition of ensuring image quality, and meet the requirements of real-time decoding.