摘要
为提高ICP匹配算法中k近邻搜索的存储和计算效率,本文分析总结了几种树结构k近邻搜索算法,利用模拟和实测数据实验对比研究了它们对ICP匹配结果的影响。实验结果表明,几种算法的拉入范围相同,匹配精度差异较小,主要差异在于搜索效率不同。其中,主轴搜索树k近邻算法的存储结构较优,近似搜索策略的计算效率较高,使得基于主轴树近似搜索的匹配效率最高。
In order to improve the storage and computation efficiency of ICP matching algorithm, k Nearest Neighbor searching algorithms（KNN） based on several tree structures were summarized in the paper. Then the influences of the KNN methods on matching accuracy were compared and analyzed with simulated and measured data. Experimental results demonstrated that the pull-in-range aspects of the methods are the same and the matching accuracies have little difference, while the main difference exists in the search efficiency. Among these methods, the Principal Axis Tree KNN searching method has superior storage structure, and the Approximate Search Strategy has higher computational efficiency, which makes the optimal matching efficiency of Approximate Principal Axis Tree.
出处
《测绘科学》
CSCD
北大核心
2014年第4期152-155,共4页
Science of Surveying and Mapping
基金
国家自然科学基金项目(41201440/D010702)
四川省教育厅资助科研项目(12ZB012)
四川省科技支撑计划项目(2013FZ0021)
关键词
匹配
ICP
K近邻
K-D树
主轴树
正交树
matching
ICP
k Nearest Neighbor
k-d tree
principal axis tree
orthogonal tree
