一种特征量化赋权的配电线点云自动提取方法

Automatic Extraction Method of Power Line Based on Point Cloud Feature Quantification and Weighting

针对配电网电力线架设低、线路与植被、构筑物等地物混淆,大规模线路高精度、高效率提取难度大的问题,提出了基于点云多特征量化赋权的电力线提取方法。该方法可根据场景内复杂地物的空间分布特点,选择地物显著区分特征,并在兼顾特征变异性和相关性的基础上量化特征区分电力线的能力,实现电力线与周围地物的分离。选用三景不同规模、地形复杂性的配电网点云数据验证该方法的有效性,实验结果表明电力线提取精度可达97.6%。

Aiming at the problems of low power line erection in distribution network,confusion between lines and vegetation,structures and oth-er ground objects,and difficulty in extracting large-scale lines with high precision and efficiency,we proposed a power line extraction method based on point cloud multi-feature quantization and weighting.According to the spatial distribution characteristics of ground objects in the scene,the method selected the features that distinguish the ground objects,and quantified the ability of features to distinguish the power lines based on taking into account the variability and correlation of the features,realizing the separation of power lines and surrounding ground objects.We se-lected point cloud data of power distribution network with different scales and terrain complexity in three scenes to verify the method's validity.The experimental result shows that the power line extraction accuracy can reach 97.6%.