• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2011, Vol. 33 ›› Issue (8): 158-161.

• 论文 • 上一篇    下一篇

AVS-M帧间预测模式快速选择算法的研究

杨丹,姚伟新,周鸣争   

  1. (安徽工程大学,安徽 芜湖 241000)
  • 收稿日期:2010-10-29 修回日期:2011-01-15 出版日期:2011-08-25 发布日期:2011-08-25
  • 作者简介:杨丹(1973),男,湖南长沙人,硕士,讲师,研究方向为模式识别、视频编码和图形图像处理。姚伟新(1977),女,辽宁阜新人,硕士,讲师,研究方向为模式识别、视频编码和图形图像处理。周鸣争(1958),男,安徽枞阳人,教授,研究方向为模式识别、视频编码和图形图像处理。
  • 基金资助:

    安徽工程大学青年科研基金(2008YQ040)

Research on Fast InterPrediction Mode Decision for AVS-M

YANG Dan,YAO Weixin,ZHOU Mingzheng   

  1. (Anhui Polytechnic University,Wuhu 241000,China)
  • Received:2010-10-29 Revised:2011-01-15 Online:2011-08-25 Published:2011-08-25

摘要:

AVS-M是我国拥有自主知识产权的数字音视频编码系列标准的移动视频标准。AVS-M采用率失真优化(RDO)技术进行帧间模式的选择,增加编码过程中宏块进行模式选择的计算复杂度。本文提出一种快速的帧间模式选择算法。该算法首先对SKIP模式提前判决,再采用二分搜索法对帧间模式的选择,对于宏块层的模式划分采用MAD(平均绝对差值)方法,对亚宏块层的模式利用视频序列的时域、空域相关性来划分。仿真结果表明,该方法保持相同视频质量的情况下,提高了编码速度。

关键词: AVS-M, 帧间预测, 模式选择

Abstract:

AVS-M is a video coding standard for mobile application,which is part of the domestic Audio Videocoding Standard(AVS)series standards. AVS-M employs rate distortion optimization(RDO) techniques in interprediction mode decision,but it increases the computation complexity of mode decision procedure during the coding process.The paper proposes a fast interprediction mode decision algorithm. Firstly, early skip mode decision is employed, Secondly, inter mode decision based on binary search methods,division of macro block level makes use of the MAD(Mean Absolute Differences) method, and according to the pacetemporal correlation, the sub macro block is divided. The experiment results show that the algorithm can reduce the encoding time while maintaining similar video quality.

Key words: AVS-M;interframe prediction;mode decision