首页    期刊浏览 2024年11月26日 星期二
登录注册

文章基本信息

  • 标题:Fast Intra Prediction Based on the Texture Feature for H.264/AVC Encoding
  • 本地全文:下载
  • 作者:Ping Wang ; Yichen Chong ; Liejun Wang
  • 期刊名称:Journal of Software Engineering
  • 印刷版ISSN:1819-4311
  • 电子版ISSN:2152-0941
  • 出版年度:2015
  • 卷号:9
  • 期号:1
  • 页码:66-76
  • DOI:10.3923/jse.2015.66.76
  • 出版社:Academic Journals Inc., USA
  • 摘要:Intra prediction in H.264/AVC high profile supports rich block types and prediction modes to achieve high coding efficiency. In order to obtain the best block type and prediction mode, the full search algorithm requires a lot of calculations. This study aims to reduce the computational complexity of intra prediction by choosing both appropriate block types and a few candidate modes. In the proposed algorithm, each macroblock is divided into sixteen 4x4 sub blocks and edge strength and direction of each sub block are obtained based on the edge detection templates. Then the mean absolute deviation of edge strengths among sixteen sub blocks is used to measure the smoothness degree of the macroblock and select the appropriate block types. For the selected block types, the candidate modes can be further chosen according to the edge direction of the block. Experimental results show that the proposed fast algorithm can achieve about 74% computational time saving on average for encoding the all intra-frame sequence with negligible loss of coding performance compared with the original full search algorithm.
国家哲学社会科学文献中心版权所有