作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 专栏 • 上一篇    下一篇

一种HEVC 帧内预测快速算法

张新晨 1,肖秀秀 1,赵 娅 1,黎 伟 2,江 昊 2,肖进胜 2   

  1. (1. 华中师范大学物理科学与技术学院,武汉430079; 2. 武汉大学电子信息学院,武汉430072)
  • 收稿日期:2013-09-03 出版日期:2014-10-15 发布日期:2014-10-13
  • 作者简介:张新晨(1977 - ),男,副教授、博士,主研方向:视频编码,多媒体通信;肖秀秀,硕士研究生;赵 娅,本科生;黎 伟,硕士研究 生;江 昊,教授、博士;肖进胜,副教授、博士。
  • 基金资助:

    国家自然科学基金资助项目(61371126);教育部-中国移动科研基金资助项目(MCM20122041)。

A Fast Intra-frame Prediction Algorithm for HEVC

ZHANG Xin-chen 1,XIAO Xiu-xiu 1,ZHAO Ya 1,LI Wei 2,JIANG Hao 2,XIAO Jin-sheng 2   

  1. (1. College of Physics Science and Technology,Central China Normal University,Wuhan 430079,China; 2. School of Electronic and Information,Wuhan University,Wuhan 430072,China)
  • Received:2013-09-03 Online:2014-10-15 Published:2014-10-13

摘要:

HEVC 作为新一代的视频编码标准,比现有H. 264 标准的压缩效率提高近一倍,但其存在复杂度较高的问 题。为此,针对HEVC 中帧内预测最耗时的模块,即编码单元块划分模块和帧内预测模式选择模块,提出一种适合HEVC 帧内预测的快速算法。该算法将率失真(RD)代价作为阈值参数,利用候选模式集中预测模式被选中概率 快速递减的规律,基于RD 代价进行帧内预测块划分和帧内预测模式选择。实验结果表明,该算法在相同编码质量 条件下可减少59% 的HM10. 0 帧内预测模块复杂度,相应比特率的增加幅度小于1. 34% 。

关键词: 视频编码, HEVC 标准, 帧内预测, 块划分, 模式选择

Abstract:

As a new generation video coding standard,HEVC has more compression efficiency than the existing video coding H. 264 standard,but it has higher complexity. Aiming at this problem,after brief analysis of the intra prediction algorithm,a fast algorithm based on Rate-distortion (RD) cost of intra prediction block dividing and the intra-frame prediction mode selection fast algorithm is proposed to reduce the HEVC intra-frame prediction complexity. This algorithm takes full use of the advantages of rate distortion cost and the probability of the prediction mode is selected in the candidate pattern quick descending law. Experimental results show that complexity reduction from HM10. 0 is over 59% and stable for various sequences,which makes the proposed algorithm suitable for real-time applications. The corresponding bit rate increase is lower than 1. 34% .

Key words: video coding, HEVC standard, intra-frame prediction, block partitioning, mode decision

中图分类号: