论文部分内容阅读
针对当前机载LiDAR技术在电力巡线应用中对电力线数字模型高精度和快速重建的需求,该文提出一种高效的电力线点云分类方法。首先基于局部范围点的高程统计直方图,实现电力线点的快速粗提取;然后运用随机抽样一致性算法剔除残留的电塔点,结合点云高程统计进一步剔除绝缘子点,实现电力线点的精提取;最后利用同一垂直面内电力线点的高程分布特性,实现单根电力线点的快速提取。基于实际输电线路机载LiDAR数据的实验结果表明,该方法可实现电力线点的快速、高精度提取:粗分类后的电力线点中仅含约10%的非电力线点;精分类后约有2%的电力线点被误分为绝缘子点,最终各条电力线点的提取比率平均为98%以上。
In order to meet the current high-precision and fast reconstruction of power line digital model in power line patrol applications, this paper presents an efficient method of power line point cloud classification. Firstly, based on the elevation statistics histogram of local range points, the power line points are quickly coarsely extracted. Then, the random sampling consistency algorithm is used to remove the remaining tower points and the points cloud elevation statistics to further remove the insulator points to achieve the refinement of the power line points. Finally, using the same vertical plane power line point elevation distribution characteristics, to achieve a single power line point fast extraction. Experimental results based on real on-board LiDAR data show that this method can realize the fast and high-precision extraction of power line points: only roughly 10% of non-power line points are included in the power line points after rough classification; about 2% Of the power line points are incorrectly classified as insulator points, and the average extraction rate of each power line point is on average more than 98%.