基于机载激光点云的电网绝缘子识别方法  

Power grid insulator identification method based on airborne laser point cloud

作　　者：曾旭 陈伯建 潘磊[3] 李诚龙 江波[1] ZENG Xu;CHEN Bojian;PAN Lei;LI Chenglong;JIANG Bo(School of Air Traffic Management,Civil Aviation Flight University of China,Guanghan 618307,China;Equipment Technology Research Center,Power Science Research Institute,State Grid Fujian Electric Power Co.,Fuzhou 350007,China;School of Computer Science,Civil Aviation Flight University of China,Guanghan 618307,China)

机构地区：[1]中国民用航空飞行学院空中交通管理学院,广汉618307 [2]国网福建省电力有限公司电力科学研究院设备技术研究中心,福州350007 [3]中国民用航空飞行学院计算机学院,广汉618307

出　　处：《激光技术》2023年第1期80-86,共7页Laser Technology

基　　金：四川省科技厅科技计划资助项目(21RKX0103);民航飞行技术与飞行安全重点实验室开放基金资助项目(FZ2021KF13);中央高校基本科研业务费重点资助项目(ZJ2021-03);民航教育人才类项目(0252103);四川省大学生创新创业训练计划资助项目(S202110624212)。

摘　　要：为了解决无人机机载激光雷达采集到的点云数据存在密度高但分布不均匀的现象,以及绝缘子表面纹理信息不全等问题,提出了一种基于机载激光点云的电网绝缘子识别方法。首先分析杆塔中不同部位的强度值直方图,用强度值滤波剔除大部分的杆身点云;然后采用主成分分析法计算局部点云特征值,根据特征值构建的局部熵函数和空间分布特性删除冗余的平坦区域点云,并通过栅格修补的方法避免出现点云空洞;最后针对传统采样一致性初始配准(SAC-IA)算法精度低和速度慢的问题,通过增加采样点对的距离约束关系和自适应调整参数改进SAC-IA算法完成绝缘子的位姿估计。结果表明,该方法能正确高效地识别杆塔中的绝缘子,运行时间大幅减少,提取正确率达到95.16%。该研究在无人机自主巡检航线规划中具有良好的应用价值。To solve the problems of high density but uneven distribution of point cloud data collected by ummanned aerial vehicle(UAV) airborne lidar and incomplete information on the surface texture of insulators, a power grid insulator identification method based on airborne laser point cloud was proposed. Firstly, the histogram of the intensity value of different parts of the tower was analyzed, and the intensity value filter was used to remove most of the tower body point cloud;the principal component analysis method was then used to calculate the local point cloud eigenvalues. The local entropy function and spatial distribution characteristics based on the eigenvalues delete redundant flat area point clouds was built. Grid patching was used to avoid point cloud holes;finally, to solve the problem of low accuracy and slow speed of the traditional sample consensus initial alignment(SAC-IA) algorithm, the SAC-IA algorithm was improved to complete the pose estimation of the insulator by increasing the distance constraint relationship of the sampling point pair and adaptive adjustment parameters. The experimental results show that the insulators in the tower can be identified accurately and efficiently by using this method. And the running time is greatly reduced, and the extraction accuracy rate reaches 95.16%, which has good application value in UAV autonomous inspection route planning.

关 键 词：激光技术 绝缘子识别 无人机 采样一致性初始配准算法 强度值滤波 主成分分析法 

分 类 号：TN249[电子电信—物理电子学] TN958.98