大规模三维点云快速拾取技术研究

Study on quick picking technology for massive 3D point cloud

为三维点云处理系统点云查询与交互编辑功能的实现,在系统总结当前计算机三维图形拾取主要方法的基础上,提出三维点云拾取基本方法。针对实际LiDAR(激光雷达)点云处理中往往为大规模点云数据,通过层次包围盒引入四叉树,提出了基于四叉树的大规模三维点云快速拾取系列算法,并从提高四叉树构建速度、降低四叉树内存占用角度,采取有效策略,使得算法整体效率得到进一步优化,实验结果表明算法在大规模三维点云拾取速度和精度上均达到了很好的效果。

To realize such functions as inquiring and interactive editing for point cloud in 3D point cloud processing system,a method for picking up 3D point cloud is proposed after systematically summarized the current main methods of picking for 3D compute graphics.To solve the problem of massive point cloud data in actual LiDAR（light detection and ranging） point cloud processing system,a series of picking algorithms for 3D point cloud based on quadtree are proposed.Meanwhile,in order to optimize the overall efficiency of the algorithms,some effective strategies concerning accelerating quadtree building and lower its memory using are adopted.The experimented results show that the algorithms achieve wonderful feedbacks both in speed and accuracy in picking for massive 3D point cloud.