基于局部限定搜索区域的特征匹配算法

Feature matching algorithm based on partial limitation search region

提出一种局部限定搜索区域的特征匹配算法,将空间约束与局部描述符结合起来。该算法在ASIFT算法基础之上,针对在特征匹配阶段直接去除一对多、多对一的特征点的缺陷做出了改进。由于这些被去掉的特征点中有很多是可以得到正确匹配的,导致获得到的匹配点对少了很多,通过在已经匹配的点对的周围限定区域内寻找出新的未曾匹配的点对,最终达到提升正确匹配数量的目标。经实验验证,所提出的局部限定搜索区域的特征匹配算法相比于ASIFT算法能大量增加特征匹配点的数量。

This paper proposes a feature matching algorithm for partial limitation search regions, which combines spatial constraints with local descriptors. Based on the ASIFT algorithm, this paper improves the defects of directly removing one-to-many and many-to-one feature points in the feature matching phase. Since many of these removed feature points can be properly matched, resulting match points less. This paper will find a new unmatched pair of points within the surrounding area of the matching pairs, and fi-nally achieve the goal of raising the correct number of matches. Experiments show that the feature matching algorithm proposed in this paper can greatly increase the number of feature matching compared with the ASIFT algorithm.