Adaptive intra mode decision for HEVC based on texture characteristics and multiple reference lines
Adaptive intra mode decision for HEVC based on texture characteristics and multiple reference lines
复制标题
基于纹理特征和多参考线的 HEVC 自适应帧内模式决策
DOI：
10.1007/s11042-018-6001-x
复制
发表时间：
2019
影响因子：
3.6
通讯作者：
Li Xi
中科院分区：
文献类型：
--
作者：
Tian Rui;Zhang Yongfei;Duan Miyi;Li Xi
高效视频编码（HEVC）标准被设计为与广泛使用的H. 264/AVC标准相比实现显著提高的编码效率。这一成就得益于高清和超高清视频应用的日益普及。然而，这是以编码器复杂度的显著增加为代价的，特别是在帧内编码中。为了增强帧内编码性能，在HEVC中采用35种帧内预测模式的集合。为了在保持编码性能的同时降低帧内预测的复杂度，提出了一种基于纹理特征和多参考线的HEVC帧内编码自适应快速模式选择算法。首先，我们利用像素值偏差（PVD），以获得占主导地位的预测单元（PU）的纹理方向和预测的纹理预测方向候选集的所有35帧内预测模式的基础上，适当考虑纹理复杂度和PU的大小。第二，一个自适应的多参考线为基础的帧内预测方案将利用分类策略，以提高编码效率。第三，将利用两个候选模式的成本之间观察到的关系，以提高预测的效率。实验结果表明，该算法平均节省了20.45%的帧内编码时间，而不会引起显着的性能下降，并通过实现更好的RD性能与近似的编码时间节省了最先进的帧内模式决策算法。
The High Efficiency Video Coding (HEVC) standard was designed to achieve significantly improved coding efficiency compared with the widespread use of H.264/AVC standards. This achievement was motivated by the ever-increasing popularity of high-definition and ultra-HD video application. However, this comes at the expense of a significant increase in encoder complexity, especially in intra-frame coding. To enhance the intra coding performance, a set of 35 intra prediction modes is adopted in HEVC. To reduce the complexity of intra prediction while maintaining the coding performance, an adaptive fast mode decision algorithm for HEVC intra coding based on texture characteristics and multiple reference lines is proposed in this paper. First, we take advantage of pixel values deviation (PVD) to obtain dominate texture direction of prediction unit (PU) and predict the texture prediction direction candidate set out of all 35 intra prediction modes based on texture direction with due consideration of texture complexity and PU size. Second, an adaptive multiple reference line-based intra prediction scheme will be utilized with classification strategy to improve coding efficiency. Third, the relation observed between the costs of two candidate modes will be exploited to improve the efficiency of prediction. Experimental results demonstrate that the proposed algorithm saves 20.45% intra encoding time on average without incurring noticeable performance degradation and outperforms the state-of-the-art intra mode decision algorithms by achieving a better RD performance with approximate encoding time saving.