为了提升高效视频编码(HEVC)帧内预测编码部分的编码效率,提出了一种HEVC帧内预测编码编码单元(CU)结构快速选择算法。算法通过对CTU(coding tree unit)四叉树结构的遍历过程进行优化,设计了两种不同的最优CU结构快速决策算法,分别从最大划分深度和最小划分深度开始遍历,并在每一步遍历之前,判断是否提前终止遍历操作。同时,在对每个CTU进行求解时,依据其纹理复杂度和当前编码状态,从两种算法中选择出最优快速决策算法对其进行求解。在HM 15.0的基础上实现了提出的快速选择算法。实验结果表明,本文算法能够在保证编码性能的同时,降低31.14%的编码时间,提高了HEVC的编码效率。
As the latest video compression standard,high efficiency video coding (HEVC) adopts a lot of novel technologies, which can improve the coding performance considerably. However, the computational complexity of HEVC also increases rapidly,especially the complexity coding tree unit (CTU) structure selection in the intra coding, which extends the encoding time seriously. In order to improve the encoding efficiency of HEVC,we propose a fast coding unit (CU) partitioning decision method for HEVC intra coding. In the proposed method,two kinds of CU partitioning structure fast decision algorithms are de- signed based on optimizing the traverse order in CTU structure selection. The traverse operations in them start from the maximum and minimum depths of CU partitioning structure respectively, and the early determination technology is also adopted. In addition, by analyzing the texture features of CTU and the coding parameters ~ the optimal fast decision algorithm can be selected from the proposed partitioning algorithms to find the optimal CU partitioning structure before checking every CTU. Based on HM 15. 0, we implement the proposed method. The experimental results show that this method can decrease the encoding time by 31.14% with minimal loss in coding performance, which means the proposed method can improve the encoding efficiency of HEVC.