机载LiDAR数据中电力线的自动提取与重建

Automatic extraction and reconstruction of powerlines from airborne LiDAR point clouds

为实现电力线走廊更加有效地巡检,本文设计了一套LiDAR点云数据中电力线自动提取与重建的方法。首先,利用改进的渐进形态学滤波剔除地面点,通过高差阈值与高程离散度分割,实现电力线点粗提取;然后,借助RANSAC直线检测,得到电力线直线模型,依靠密度检测,实现单根电力线点云精确聚类;此外,利用k-means算法完成分裂导线束间归类;最后,进行二次多项式限制的最小二乘拟合,生成电力线曲线模型。试验结果表明,使用该方法电力线点云提取的正确率达98%以上,非电力线点云误判率低至1%左右,电力线直线模型拟合误差在5 cm以下,曲线模型拟合误差在3 cm以下,完全满足实际工程需求。

In order to realize more effective inspection of powerline corridor,this paper designs a method for automatic extraction and reconstruction of powerline from LiDAR point cloud data.The improved progressive morphology filtering is used to eliminate the ground point.It achieves rough extraction of powerline points through the segmentation based on the height difference threshold and elevation dispersion.The powerline linear model is obtained through RANSAC line inspection,relying on density detection,precise clustering of single powerline points is achieved.It complete the bundled conductors classification with the k-means algorithm.Finally,a least square fitting of the quadratic polynomial is performed to generate a powerline curve model.Experiments show that the correct rate of powerline point cloud extraction is over 98%,the false positive rate of non-powerline point cloud is as low as 1%,the fitting error of powerline linear model is below 5 cm and the curve model is below 3 cm.Fully meet the actual engineering needs.