摘要
在充分考虑了各种图像序列的运动特性的基础上,提出一种基于矢量预测和多方向梯度下降搜索(MDGDS)的运动估计算法。该算法首先利用运动矢量的时间和空间上的相关性来得到预测矢量的搜索起始点,再通过使用自适应阈值来判断当前块的运动类型,以此智能地选择不同的搜索策略。实验结果表明,与FS、UMHexagonS、简化UMHS以及EPZS等传统算法相比,该算法能在保证精度的同时节约大量的编码时间。
On the basis of fully considering the motion characteristics of various video series, a motion estimation algorithm based on vector prediction and multi-direction gradient descent search is proposed. First the algorithm utilizes the temporal and spatial correlations of the motion vectors to obtain the search starting point of a predictive vector, then it determines the motion type of the current block by using adaptive thresholds, so as to select the different search strategy intelligently. The experiment results show that the algorithm can save more encoding time than the traditional FS, UMHexagonS, simplified UMHS and EPZS algorithms with the same accuracy.
出处
《计算机时代》
2010年第3期4-6,共3页
Computer Era
基金
浙江省自然科学基金项目(Y106657)
浙江省钱江人才科技计划项目(2006R10010)
关键词
运动估计
块匹配
矢量预测
自适应阈值
motion estimation
block matching
vector prediction
adaptive threshold
