基于无人机载LiDAR的输电线路树障单木识别

刘凤莲; 曹永兴; 高润明; 郭裕钧; 刘凯

刘凤莲,曹永兴,高润明,郭裕钧,刘凯.基于无人机载LiDAR的输电线路树障单木识别[J].电测与仪表,2023,60(9):7-13.
Liu Fenglian,Cao Yongxing,Gao Runming,Guo Yujun,Liu Kai.Recognition of tree barriers on transmission lines based on UAV-borne LiDAR[J].Electrical Measurement & Instrumentation,2023,60(9):7-13.

基于无人机载LiDAR的输电线路树障单木识别

Recognition of tree barriers on transmission lines based on UAV-borne LiDAR

DOI：10.19753/j.issn1001-1390.2023.09.002

中文关键词: 树障激光雷达点云特征种类识别

英文关键词:tree barriers, LiDAR, point cloud feature, type recognition

基金项目:国家电网有限公司科技项目（521997170013）

作者中文名	作者英文名	单位
刘凤莲	Liu Fenglian	国网四川省电力公司电力科学研究院，成都 610072
曹永兴	Cao Yongxing	国网四川省电力公司电力科学研究院，成都 610072
高润明	Gao Runming	西南交通大学电气工程学院，成都 611756
郭裕钧	Guo Yujun	西南交通大学电气工程学院，成都 611756
刘凯	Liu Kai	西南交通大学电气工程学院，成都 611756

摘要点击次数: 1465

中文摘要:

树障是高压输电线路在复杂山区植被茂密区域运行所面临的主要安全威胁之一，不同树种生长周期不同，一定时间内的树障风险也不同。为了大范围准确识别林区的树木种类，文中提出了一种基于机载雷达测量技术的树木种类快速识别方法。利用机载雷达对输电线路地区地面进行快速点云数据获取，并且预处理数据得到单棵树木的冠层点云；建立冠层的空间属性点云特征量，包括树冠高度、树冠体积、树冠点云密度、冠层激光反射强度以及树冠形貌特征；根据树木的空间点云特征建立树木的种类K均值聚类识别模型，并与树木光谱特征分类结果进行对比。结果表明：对于该地区生长的树木，相对于光谱特征下的树种分类，5种空间点云特征具有良好的识别效果，最终建立的树木种类K均值聚类识别模型对于验证数据的准确率达到了85.9%，Kappa系数为0.812。输电线路下方植被种类的快速识别对于树障风险评估和预警具有重要意义。

英文摘要:

Tree barriers are one of the main safety threats faced by high-voltage transmission lines operating in dense mountainous areas with dense vegetation. Different tree species have different growth cycles and different risks of tree barriers within a certain period of time. In order to accurately identify the types of trees in the forest area on a large scale, this paper proposes a rapid recognition method for tree types based on airborne LiDAR measurement technology. The airborne LiDAR is used to quickly obtain point cloud data on the ground of the transmission line area, and preprocess the data to obtain the canopy point cloud of a single tree. The point cloud feature quantities of canopy spatial attribute is established, including canopy height, canopy volume, canopy point cloud density, canopy laser reflection intensity, and canopy topography features. A tree-type K-means clustering recognition model is established based on the spatial point cloud characteristics of trees, and compared with the classification results of the spectral characteristics of trees. The results show that for the trees growing in this area, the five spatial point cloud features have a good recognition effect compared to the tree species classification under the spectral characteristics. The final establishment of the K-means clustering recognition model for trees has an accuracy rate of 89% and a kappa of 0.812 for the verification data. Rapid identification of vegetation species under transmission lines is of great significance to the risk assessment and early warning of tree barriers.

查看全文查看/发表评论下载PDF阅读器

关闭