3D-HEVC合并模式快速判决方法研究

基于高效视频编码的3D视频编码（3D-HEVC）是目前正在研究的新一代3D视频编码标准。为降低3D-HEVC中模式选择的计算复杂度，根据非独立视点纹理图中合并模式采用率高的特点，该文提出了一种3D-HEVC合并模式快速判决方法。在B帧中，分析了当前编码单元（CU）与视点方向参考帧中参考块间编码模式的相关性；在P帧中，分析了位于相邻划分深度的CU间编码模式的相关性。根据分析的视点间和划分深度间的相关性设计快速判决条件，预判采用合并/合并-跳过模式编码的CU，判别出的CU在模式选择过程中只检查相关的候选预测模式，从而降低计算复杂度。实验结果表明，与3D-HEVC原始算法相比，该文算法能够在率失真性能损失很小的前提下，平均节省11.2%的总编码时间和25.4%的非独立视点纹理图的编码时间。 

Research on Fast Merge Mode Decision Method for 3D-HEVC

High Efficiency Video Coding based 3D Video Coding (3D-HEVC) is a novel 3D video coding standard under research. In the test model of 3D-HEVC, due to the exhaustive checks of prediction modes for coding units (CU) derived from quad-tree partitioning structure, the mode decision process brings significantly high computational complexity. To reduce the complexity of mode decision, a fast merge mode decision method was proposed based on the high selection rate of merge mode in dependent texture views. For B frame, the correlation between coding modes of the current CU and inter-view reference blocks was analyzed. For P frame, the correlation between coding modes of CUs located in adjacent depth levels was analyzed. Based on the correlations, conditions of fast merge mode decision and fast skip mode decision were designed. When the conditions were satisfied, merge mode coded CUs and skip mode coded CUs were early determined. For the early determined CUs, only relevant candidate prediction modes were checked and other prediction modes were skipped to reduce computational complexity. Experimental results show that average 11.2% of total encoding time saving and 25.4% of encoding time saving for dependent texture views can be achieved with negligible rate distortion performance loss compared with the original method in 3D-HEVC.