摘要： 为更加有效地提高运动估计速度，提出一种自适应动态搜索范围运动估计算法，从后续快速运动估计算法的运动矢量预测集中自适应地选择与当前编码块相关性最强的运动矢量预测值作为搜索范围的中心点，根据预测集中运动矢量预测值的大小、方向自适应地决定水平、垂直及正负方向的非对称搜索范围。将该算法融合到UMHexagonS 和 FFS算法中进行广泛的实验测试，结果表明其能在基本保持重建图像质量的同时，至少分别减少运动估计运算量的22.13%和76.57%。

关键词: 视频编码, 运动估计, 运动矢量, 动态搜索范围

Abstract: To accelerate the motion search process, this paper proposes a novel adaptive dynamic search range algorithm. It selectes dynamically the predictive Motion Vector(MV), whose relevant referenced block is most correlative with the current block, in candidate MV set of backward fast motion estimation algorithm as the center of search range. It determines adaptively the unsymmetrical search range in negative/positive direction of horizontal/vertical axis according to the predictive MV set to reduce the search points. Extensive experiments show that this proposed scheme, combined with UMHexagonS and FFS, can achieve more than 22.13% and 76.57% computational reduction of motion estimation respectively, compared with the previous algorithms, while maintaining almost the same quality of reconstructed pictures.

Key words: vedio coding, motion estimation, Motion Vector(MV), dynamic search range

中图分类号: 

• TN919.8