基于机载LiDAR点云的电力线自动提取

Low voltage power line extraction based on airborne LiDAR

摘要:

基于机载激光雷达数据提出了一种在林区中电力线自动提取方法,该方法是基于统计分析和二值化图像处理技术设计。首先采用高度阈值,分离出电力线候选数据集,并采用一组标准(例如,高度标准,密度标准和直方图阈值)来对候选集进行统计分析,选择电力线的候选点。然后将候选点转化为二值化图像,并进行形态学优化,采用基于图像的处理技术,利用渐进概率霍夫线性变换对图像进行直线分割。最后将分割出来的电力线二值化图像转换成三维点云,并利用区域增长精细化提取电力线点云。使用不同林区环境下的4组机载激光雷达数据进行实验,实验结果表明,算法在林区环境下能够完整地提取出电力线,且电力线分类精度较高,对于电力巡线具有较高的利用价值。

Abstract:

This paper presents an automatic power line extraction method based on LIDAR data in forest. The method is based on statistical analysis and binary image processing technology. First, the height threshold is used to separate the power line candidate data, and a set of criteria (for example, height criteria, density criteria and histogram threshold) are used to statistically analyze the candidate set and select the candidate point of the power line. Then the candidate points are transformed into binary image, and the morphological optimization is carried out. The image segmentation is carried out by using the probability-based Hough linear transformation. Finally, the segmentation of the power line image were transformed into a 3D points cloud, and the power line point cloud was refined by using region growth. Experiments were carried out using 4 sets of LiDAR data in different forest environments. The results show that the proposed algorithm can extract the power line completely in the forest environment, and the power line classification accuracy is higher, which has higher utilization value for the power line inspection.