Edge information based fast selection algorithm for intra prediction of HEVC
Edge information based fast selection algorithm for intra prediction of HEVC
复制标题
DOI：
10.1109/apccas.2014.7032708
复制
发表时间：
2014-11
期刊：
影响因子：
--
通讯作者：
Wen Shi;Xiantao Jiang;Tian Song;T. Shimamoto
中科院分区：
文献类型：
--
作者：
Wen Shi;Xiantao Jiang;Tian Song;T. Shimamoto
与先前的视频编码标准相比，高效视频编码（HEVC）实现了更好的编码效率。为了提高帧内编码效率，HEVC采用灵活的编码单元结构，如从4×4到64×64定义的预测单元（PU）大小和大量的预测模式。HEVC中帧内预测的这些新特性提高了编码效率，同时也带来了较高的计算复杂度。提出了一种HEVC帧内预测的快速算法。设计了一种基于边缘信息的快速图像预处理阶段。根据边缘检测的结果，有效地确定了合理的PU尺寸。基于边缘信息的帧内预测模式选择优化也有助于实现降低计算复杂度的目标。与原始HEVC相比，复杂度降低了27%以上，同时，BD-Rate的增加和BD-PSNR的下降几乎可以忽略不计。
High efficiency video coding (HEVC) achieves a better coding efficiency compared with previous video coding standards. To enhance the intra coding efficiency, HEVC employs flexible coding unit structure as prediction unit (PU) sizes which are defined from 4×4 to 64×64 and plenty of prediction modes. These new features of intra prediction in HEVC improves coding efficiency, meanwhile, brings high computational complexity. This paper presents a fast algorithm for intra prediction of HEVC. An edge information based fast image preprocessing stage is designed. Based on the result of edge detection, reasonable PU size is decided efficiently. Intra prediction mode selection optimization based on edge information also helps to achieve the goal of computational complexity reduction. Compared to the original HEVC, the result of the complexity reduction is over 27%, in the meantime, the increased BD-Rate and declining BD-PSNR is almost negligible.