点云数据的k近邻快速建立改进算法
安雁艳;杨秋翔;冯欣悦;范建华;杨剑
【期刊名称】《计算机工程与设计》 【年(卷),期】2014(000)012
【摘要】For the problem of bigger sub‐block space after dividing when searching point’ s nearest neighbors and extending around search range before completing search ,a fast search algorithm based on second dividing and controlled extension direction was presented .At first ,the minimum bounding box of point cloud data was divided into a set of uniform grids by using traditional algorithms .Then the size of second divided grid was figured out by overall considering grids that were not empty ,the number of nearest neighbors and the size of first divided grid .Finally ,the most possible grids were searched preferentially through direction control in the process of local search .Compared with the existing methods ,simulation results shows that the provided algorithm can reduce searching time by 20% at least .%针对点云数据最近点搜索时栅格化所得空间子块大，并且在未完全找到前搜索范围需扩展一圈的问题，提出一种基于二次栅格化和扩展方向可控的快速搜索算法。采用传统分块算法一次栅格化数据空间；综合考虑非空栅格、最近点数目及一次划分边长，计算二次栅格化的边长；在局部搜索过程中控制扩展方向，优先在最有可能出现的栅格中进行搜索。实验结果表明，与现存的方法相比，该算法在搜索时间上至少减少了20％。
点云数据的k近邻快速建立改进算法
