激光技术, 2023, 47 (2): 233, 网络出版: 2023-04-12  

一种地下电缆点云自动提取分割算法

An automatic extraction and segmentation algorithm of underground cable point cloud
作者单位
国网河北省电力有限公司经济技术研究院, 石家庄 050000
摘要
为了解决目前地下电缆隧道点云中电缆支架与电缆分离困难、电缆点需要人工提取的问题, 提出一种基于区域圆柱面拟合与抗差自适应Kalman滤波的地下电缆点云自动提取分割算法。首先基于电缆局部呈圆柱的形状特征, 将区域点云进行圆柱面拟合以确定电缆初始区域的中心轴线与半径; 再将初始区域电缆的轴线作为区域电缆走向, 结合抗差自适应Kalman滤波算法对电缆中轴线进行延长估计, 得到单条完整电缆中心轴线, 最终实现电缆点云单条分割。结果表明, 电缆初始区域单位权中误差在0.015 m以内, 能够有效区分电缆点与其它类别点; 电缆中轴线延长在地下电缆点云存在缺点、噪点情况仍保持稳健估计, 具有较好的鲁棒性。该方法有效提升了地下电缆提取的准确性和可靠性。
Abstract
In order to solve the problems that it is difficult to separate the cable support from the cable in the point cloud of underground cable tunnel and the cable points need to be extracted manually, an automatic extraction and segmentation algorithm of underground cable point cloud based on regional cylindrical surface fitting and robust adaptive Kalman filter was proposed. Firstly, the radius of the cable area was determined based on the shape of the cylindrical axis; Then, the axis of the initial regional cable was taken as the direction of the regional cable, and the extension of the central axis of the cable was estimated combined with the robust adaptive Kalman filter algorithm to obtain the central axis of a single complete cable. Finally, the single segmentation of the cable point cloud was realized. The results show that the mean square error of unit weight in the initial area of cable is less than 0.015 m, which can effectively distinguish cable points from other types of points; The cable central axis extension has disadvantages in the underground cable point cloud, and the noise still maintains a robust estimation, which has good robustness. This method effectively improves the accuracy and reliability of underground cable extraction.

李军阔, 刘建, 李光毅, 任雨, 邵华. 一种地下电缆点云自动提取分割算法[J]. 激光技术, 2023, 47(2): 233. LI Junkuo, LIU Jian, LI Guangyi, REN Yu, SHAO Hua. An automatic extraction and segmentation algorithm of underground cable point cloud[J]. Laser Technology, 2023, 47(2): 233.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!