摘要
基于时运动矢量概率分布的方向性中心偏移特性和起点预测搜索技术的研究,提出了一种起点预测的十字形快速运动估计算法(IPPAFA),在保持搜索精度的同时有效降低了计算复杂度。实验结果表明,在图像质量几乎不下降的情况下,与钻石搜索算法(DS)和运动矢量场自适应搜索算法(MVFAST)相比,该算法在速度和准确性方面都优于传统的快速运动估计算法。
Based on the research of probability directional centre - based motion vector distribution and predictive search, this paper proposes an initial search point predictive across - shape fast motion estimation algorithm （IPPAFA）. The new algorithm can effectively reduce computational complexity while keeping search accuracy. Experimental results show that the proposed algorithm excels other traditional algorithms in speed and veracity, with no obvious loss of video quality in contrast to DS and MVFAST.
出处
《电子科技》
2007年第7期36-39,共4页
Electronic Science and Technology
关键词
运动估计
十字形搜索
起点预测
IPPAFA
motion estimation
acrossshape search
initial search point prediction
IPPAFA
