一种改进的基于法矢方向调整的平面点云分割方法

Planar Point Cloud Segmentation Based on the Weighted Average of Adjusted Normal Vector

智慧城市的蓬勃发展和快速推广,使得三维建模成为当前热点研究方向。平面点云分割是三维点云建模中数据处理的关键环节。该文提出一种三角面片法向量方向调整方法,通过后续对邻近法向量进行加权平均估算实现平面点云的分割。首先采用八叉树的空间划分方法将无序的点云建立索引,利用K紧邻搜索获取参考点的K个邻近点,然后将该局部点构建不规则三角网并且得到包含参考点的所有三角面片以及三角面片的法向量,并通过将三维点投影到二维平面,利用平面三角形两边向量叉乘的方法,判断并调整各三角面片的顶点排列顺序,使三角面片的法向量一致化,最后对包含参考点的所有三角面片的法向量加权平均,估算参考点的法向量,根据点云法矢一致、共面的原则将平面点云分割出来。以徕卡Scanstation 2型扫描仪获取点云数据,对该方法进行检验,结果显示其能较好地实现对点云法矢量方向的调整与估算,并对平面点云数据进行分割提取。

With the vigorous development and rapid promotion of wisdom city, 3D modeling is becoming the current hot research direction. Planar point cloud segmentation is the key part of three-dimensional point cloud modeling. In this paper, a kind of ad- justment method is put forward in the normal vector direction of triangles, then weighted average of the adjacent normal vector is done to realize the plane segmentation of point cloud. Firstly, using Octree space division method to establish an index for the scattered point cloud, using K neighboring algorithm to obtain K neighboring points, then triangular irregular network is estab- lished, then to project the 3D point on a two-dimensional plane, according to the product of two edge vectors of triangles, the tri- angle vertex order is adjusted to make the normal vector point to the same side, and the weighted average of adjusted normal vector is used to estimate the normal vector of a point, finally the planar point is extracted based on the consistence in direction and coplanar condition. Taking the point cloud obtained by Seanstation 2 scanner as an example, the result shows that this meth- od can realize the plane segmentation of point cloud data.