An efficient fast mode decision algorithm for H.264/AVC intra prediction

Guifen Tian, Tianruo Zhang, Xianghui Wei, Takeshi Ikenaga, Satoshi Goto

研究成果: Conference contribution

6 引用 (Scopus)

抜粋

Rate distortion optimization (RDO) based spatial intra prediction is an important new feature in H.264/AVC. It chooses the best mode for each MB by exhaustively examining all possible modes. Thus, drastically increased computation complexity is involved. This paper proposes an efficient fast intra mode decision algorithm. Prewitt operator is used to extract each block's two most dominant edge directions. Based on these two directions, only few modes are chosen as candidates. In this way, computational load caused by RDO can be reduced remarkably. Experimental results show that more than 70% computation reduction can be achieved, with little loss of visual quality. Moreover, only addition and comparison operations are used, which makes proposed algorithm friendly enough to be implemented in hardware.

元の言語English
ホスト出版物のタイトルProceedings - 1st International Congress on Image and Signal Processing, CISP 2008
ページ411-415
ページ数5
DOI
出版物ステータスPublished - 2008 9 22
イベント1st International Congress on Image and Signal Processing, CISP 2008 - Sanya, Hainan, China
継続期間: 2008 5 272008 5 30

出版物シリーズ

名前Proceedings - 1st International Congress on Image and Signal Processing, CISP 2008
1

Conference

Conference1st International Congress on Image and Signal Processing, CISP 2008
China
Sanya, Hainan
期間08/5/2708/5/30

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Signal Processing

フィンガープリント An efficient fast mode decision algorithm for H.264/AVC intra prediction' の研究トピックを掘り下げます。これらはともに一意のフィンガープリントを構成します。

  • これを引用

    Tian, G., Zhang, T., Wei, X., Ikenaga, T., & Goto, S. (2008). An efficient fast mode decision algorithm for H.264/AVC intra prediction. : Proceedings - 1st International Congress on Image and Signal Processing, CISP 2008 (pp. 411-415). [4566189] (Proceedings - 1st International Congress on Image and Signal Processing, CISP 2008; 巻数 1). https://doi.org/10.1109/CISP.2008.489