摘要： 提出一种采用弧形、小十字和大菱形模板进行快速块匹配运动估计的算法NODAS，利用序列图像中运动矢量场中心偏置分布特性，将小十字模板扩展为方向性菱形模板处理中心区域的搜索，使用大菱形-弧形模板处理大运动矢量的搜索。试验结果证明，在搜索精度基本相同的情况下，NODAS算法能有效减少搜索点数，提高搜索速度，较好地应用于小运动矢量和大运动矢量的搜索中。

关键词: 块匹配算法, 运动估计, 方向性菱形搜索, 弧形搜索

Abstract: This paper presents a block motion estimation algorithm named Novel Oriented Diamond-Arc Search(NODAS) using small cross-diamond-arc search pattern. According to the characteristic of the center-based motion vector distribution for the real-world video sequence, NODAS employs small cross pattern extended oriented diamond search pattern to refine the central area search and employs diamond-arc search pattern for large motion vector. Experimental results indicate that NODAS algorithm can reduce the number of search points to improve the search speed with the similar search precision, and it does well in stationary and large motion search.

Key words: block-matching algorithm, motion estimation, oriented diamond search, arc search

中图分类号: 

• TP301.6