基于扫描方式与球面投影的点云修补方法

Point cloud repair method based on scanning mode and spherical projection

激光雷达点云数据修补是三维重建工作中的重要环节。针对现有的点云修补的插值算法无法确定空洞区域缺失点具体位置和数量的问题,提出一种简便有效的缺失点云修补方法。首先,将经过球面投影的单站激光雷达数据按照扫描方式实现结构化;然后根据旋转角和俯仰角两个方向的角度分辨率来描述缺失点云的球面位置;最后根据实际地形情况选择合理的插值方式估计高程值,得到缺失点完整的三维坐标。以单站激光雷达栅形扫描为例,实验表明本方法能找出准确数量的点云,修补后空间位置误差率仅为6%,且经过三维空间插值后点云密集平滑。利用方便获取的已知信息,可以准确得到缺失点的分布信息。修补结果真实可靠,有利于后续三维重建工作。

Lidar point cloud data patching is an important part of 3 D reconstruction.Aiming at the problem that existing interpolation algorithms of point cloud repair can not determine specific location and number of missing points in cavity area,a simple and effective missing point cloud repair method is proposed.Firstly,the single station lidar data through spherical projection is organized according to scanning mode.Then,spherical position of missing point cloud is described according to angular resolution of rotation angle and pitch angle.Finally,according to actual terrain,a reasonable interpolation method is selected to estimate the elevation value,and the complete three-dimensional coordinates of missing point cloud are obtained.Taking the grid scanning of single station lidar as an example,the experiment shows that this method can find an accurate number of point clouds,the spatial position error rate after repair is only 6%,and the point clouds are dense and smooth after three-dimensional spatial interpolation.The distribution information of missing point cloud can be accurately obtained by using easily obtained known information.The repair results are real and reliable,which is conducive to the follow-up three-dimensional reconstruction.