论文部分内容阅读
高性能视频编码(HEVC)作为最新的视频编码标准,有赖于各种新技术的引入,其中Merge模式利用部分相邻块运动信息的相关性对其的压缩性能有显著贡献,然而HEVC在显著提升编码效率的同时复杂度也有所提高,限制了它在实际中的应用。由于需要遍历多个Merge候选点使Merge模式带来一定的复杂度,为降低Merge模式的编码复杂度,本文分析了不同编码单元深度和量化系数值所对应的率失真代价值(RD cos t)变化特点,提出使用自适应阈值的方法减少Merge模式中的率失真优化计算次数,从而快速选择最优的Merge候选点。基于HM14.0的仿真结果显示,所提方法能平均减少15.38%的编码时间,而在相同客观质量下码率只增加了0.37%。
As the latest video coding standard, High Performance Video Coding (HEVC) relies on the introduction of various new technologies. Merge mode makes a significant contribution to its compression performance by utilizing the correlation of the motion information of some adjacent blocks. However, HEVC is significantly improved The complexity of encoding efficiency is also increased, limiting its practical application. Because Merge mode needs to be traversed to bring some complexity to Merge mode, in order to reduce the coding complexity of Merge mode, this paper analyzes the rate distortion cost value (RD cos t) corresponding to different coding unit depth and quantization coefficient value, The paper proposes a method of adaptive threshold to reduce the rate-distortion optimization calculation in Merge mode so that the optimal Merge candidate can be quickly selected. The simulation results based on HM14.0 show that the proposed method can reduce the coding time by 15.38% on average, while the coding rate only increases by 0.37% under the same objective quality.