共享近邻聚类算法点云面线提取

Research on extraction of point cloud data surface and line based on shared neighbor clustering algorithm

针对激光点云数据进行建筑物建模或矢量信息提取中快速识别建筑物面和棱线信息的要求,该文提出基于共享近邻聚类算法进行建筑物面和棱线的快速提取方法。首先,计算点云中每个数据点的单位法向量和点到基准面的距离,利用基于网格的共享近邻聚类算法对点云进行分类确定建筑物面点云;然后,自动判别相交平面,提取建筑物棱线,并与RANSAC算法对某建筑物面的提取结果进行比较。结果证明,该方法自动化程度高,建筑物面和棱线提取快速、准确,提取结果能够应用于三维建筑物自动建模和测绘出图。

The information of building surface and ridge line need to be quickly identified when using the laser point clouds data for building modeling or vector information extraction. A fast extraction method for building surface and ridge line based onShared Neighbor Clustering algorithm is proposed in this paper. Firstly, the unit normal vector of each data point in point cloud and the distance between point and the reference plane are calculated, and the Shared Neighbor Clustering algorithm based on the grid is used to classify the point cloud to determine the building surface point cloud. Then, the intersection plane is au- tomatically judged and the building edges are extracted, and the extraction results of a building surface based on RANSAC algorithm are compared with the extraction results of the building edges. The results show that the method has high degree of automation, the extraction of building surface and ridge line is fast and accurate, and the extraction results can be applied to 3d building automatic modeling and mapping