Implementation And Optimization Of AVS Interframe Decoding Algorithm

As a new generation of digital audio/video encoding and decoding technology standard with independent intellectual property rights in China, AVS shows lower complexity and well function in internet multimedia and high-definition digital video and other applications. Under the condition of equal picture quality, compared with traditional MPEG-2, data compression efficiency of AVS standard improved2-3times, it is equivalent to MPEG-4and AVC/H.264but reduces30%to50%than computing resources. This means it greatly saves costs of hardware and software and has an immense development prospect. At present, the decoder reference software that AVS official supplies can not reach the real-time decoding of most video images. This is not conducive to AVS’s industrialization promotion. It needs continuous reasonable improvement and effective optimization to achieve the requirements of real-time decoding for basic CIF, Dl and other format images. Therefore, under the background of increasingly widespread application of multimedia and increasing of communication pressure that internet endures, the research for efficient AVS video decoder has great significance. It is based on the above application requirements, through intensive study of AVS video decoding standard, the thesis designed and developed a PC platform-based AVS video decoder, and conducted post-optimization by using SIMD technology in Visual Studio2008. Under the premise of ensuring the quality of the decoded image, decoder’s operating speed gets significant improvement and it has practical meaning. The thesis’s main work contents and achievements are as follows:Above all, the thesis analyzes AVS Standard’s decoding system structure, core technology and bittorrent structure. Based on the further study of AVS standard reference software, it completed frame structure design of AVS video decoder and provides an implementation plan of language C.Next, it gives a detailed analysis about AVS integer inverse transform and realization of interframe prediction algorithm. Meanwhile, it completed realization of code C in AVS decoding algorithm. And the following aspects achieved C-level preliminary optimization including intraframe, interframe separate decoding, macroblock information structure’s redefinition, optimization space distribution, all-zero block prior treatment.Finally, the thesis focused mostly on study and design of AVS integer inverse transform algorithm in basis of SSE2. It analyzed and improved sample interpolation algorithm in detail. And part codes that need large amount calculation and time-consuming are optimized by using SIMD technology. Real-time decoding that based on PC platform is available in the end. Of many kinds of different resolutions’different video sequence tests’ results shows AVS video decoder the thesis designed has well performance. Ensuring decoding quality, computing speed of optimized decoder has significantly improvement. It indicates the design and optimized program are feasible and effective, and has important meaning to real-applications of AVS decoder.