针对现有电力线激光雷达点云分割方法中存在的问题,该文提出了一种采用特征空间K-means聚类的单档电力线激光雷达点云分割方法:利用电力线LiDAR点云的水平坐标信息进行直线拟合,并对LiDAR点沿直线方向进行分段;将每一段LiDAR点云投影到相应的电力线切平面(该平面垂直于拟合直线);最后使用K-means聚类方法进行投影点的聚类,且相邻的段和段之间通过投影中心点进行类别的传递和规则化。实验表明,该方法可以较好地进行单档电力线LiDAR点云分割,且对电力线根数、电力线类型、电力线空间配置结构、档距长度、点云不规则断裂等因素不敏感。 Aiming at the existing problems in point cloud segmentation of powerline Lidar, this paper proposes a point cloud segmentation method based on K-means clustering of feature space, which is based on the horizontal coordinate information of LiDAR point cloud Line fitting, and segment LiDAR points in a straight line direction; project each LiDAR point cloud to a corresponding power line slicing plane (the plane is perpendicular to the fitted line); and finally use a K-means clustering method to project a projection point Clustering, and the adjacent segments and segments through the center of projection for the type of transmission and regularization. Experiments show that this method can be used to segment LiDAR point cloud of single power line, and it is insensitive to such factors as the number of power lines, the type of power lines, the space configuration of power lines, the length of span, and the irregular rupture of point cloud.