中文摘要：

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%。

英文摘要：

Merge mode makes full use of the relation between blocks and their nei ghbors to significantly improve the compression of high efficiency video coding （HEVC）.However,greater computati on complexity is introduced compared with the previous video coding standards,which will limit its practical applications.Checking all the merge mode candidates for the best one introduces new complexity to Merge mode.To reduce the complexity of Me rge mode,first of all,this paper analyzes the characteristics of rate distortion （RD） cost when depths and quantitive parameters （QPs） of the blocks are different,then proposes an adaptive method to reduce the complexity of RD-cost compution to get the best m erge candidate.As a result, unnecessary merge mode selection can be skipped and terminated.According to th e simulation results based on HM 14.0,conclusions are obtained that the proposed method can reduce about 15.38% encoding time with only 0.37% Bjntegaard delta （BD）-rate increase,while maintaining almost the same R D performance as the original HEVC encoder.