摘要： 高效视频编码的Merge模式基于四叉树的图像编码方式,可以将多个相邻且具有相似运动情况的预测单元(PU)块合并成同一个区域并使用相同的运动参数。研究候选PU的合并决策过程,针对Merge模式所带来的高计算复杂度且合并候选过程中常进行多次不必要的判断问题,提
出一种低复杂度的候选决策算法。该算法基于时空域候选在不同情形下的分布特征,从改变候选列表选择过程的角度减少候选判断次数从而降低编码复杂度。实验结果表明,与原始算法相比,该算法的编码时间缩短、码率减少,能有效降低编码复杂度。
关键词:
高效视频编码,
帧间预测,
Merge模式,
运动矢量,
时空域候选,
候选顺序
Abstract: High Efficiency Video Coding(HEVC) Merge model based on quad-tree image coding method is proposed,which can merge several closed PU blocks that have similar motion cases into the same area,using the same motion parameters.This article studies the Merge decision-making process of the candidate PU block,for Merge mode brings high computational complexity and there is a lot of unnecessary computation in the process.It proposes a low complexity candidate decision algorithm.The algorithm is based on the distribution characteristics of spatio-temporal domain candidate and reduces the number of candidates from the perspective of changing the process of candidate list selection to reduce the complexity of encoding.Experimental results show that,compared with the original algorithm,the encoding time and bit-rate of the proposed algorithm are reduced,and the algorithm is proved to improve the rate-distortion performance while effectively reducing the coding complexity.
Key words:
High Efficiency Video Coding(HEVC),
inter prediction,
Merge mode,
motion vector,
spatio-temporal domain candidate,
candidate order
中图分类号: