向红莉 张刚

摘 要：针对帧内预测的快速算法，由于DSP架构软件顺序执行的局限性难以满足实时性要求，而FPGA以其高速的计算速度和强大的并行处理能力成为H.264和AVS编解码的理想平台。本文在FPGA平台上采用资源共享、高并行和多流水线结构实现了亮度帧内预测算法。该方法在分析AVS帧内亮度5种预测模式的基础上，将像素预测与模式判决在一个模块中完成，并且利用各模式预测的相似性，实现运算单元共享和多种模式并行执行，兼顾了处理速度和实现代价。仿真及综合结果表明该设计能够完全满足标清(704×576，30f/s)数字视频的实时处理要求。

关键词：AVS；帧内预测；模式判决；FPGA

中图分类号：TN919.8文献标识码：aDoI: 10.3969/j.issn.1003-6970.2012.02.036

Intra Frame Prediction for AVS Encoder based on FPGA

XIaNG Hong-liZHaNG Gang(College of Information Engineering Taiyuan University of Technology, Taiyuan 030024, China)

【Abstract】The fast algorithm of intra prediction is difficult to meet the real-time demand on DSP because of the limitations of sequential order of DSP software architecture; while FPGA with the high calculation speed and strong parallel processing ability has become the ideal platform for H.264 and aVS codec. In this paper, a method which applies the resource-sharing, high parallel and pipeline structure has been put foward .The method, in the analysis of 5 prediction modes of AVS intra luma and their similarity in operation, has put the pixel prediction and mode decision in a module and realized the sharing of public operation unit and concurrent execution of a variety of modes on consideration of both processing speed and realizing costs. the simulation and comprehensive results show that the design has been completely meet real-time processing requirements of the standard (704 x 576, 30 f/s) digital video.

【Key words】aVS; Intra frame Prediction;Mode Decision; fPGa

0 引 言

AVS(Audio Video Coding Standard)标准[1],是《信息技术—先进音视频编码》系列标准的简称, 是由我国自主提出的数字音视频编解码技术标准,这个标准达到了当前国际先进水平。AVS具有以下优点:(1)性能高,编码效率是MPEG2的2倍以上,与H.264的编码效率处于同一水平;(2)复杂度低,算法复杂度比H.264明显低,软硬件实现成本都低于H.264;(3)我国掌握主要知识产权,专利授权模式简单,费用低[2]。AVS视频标准制定的目标是: 适应数字电视广播、数字存储媒体、因特网流媒体、多媒体通信等应用中对运动图像压缩技术的需要，这就对编码的实时性提出了很高的要求。

帧内预测[3]是AVS编解码中很重要的部分，其原理是利用图像内部相邻像素间的相关性来降低编码码率，也就是通过减少空间冗余度来达到压缩的目的。视频编码器帧内预测部分的硬件架构设计方案一般分为两种。……