中国电力 ›› 2019, Vol. 52 ›› Issue (1): 82-87,95.DOI: 10.11930/j.issn.1004-9649.201804103

• 电网 • 上一篇    下一篇

面向航拍图像的农村配电网导线断股检测

李泊1, 陈诚2   

  1. 1. 南京农业大学 工学院, 江苏 南京 210031;
    2. 全球能源互联网研究院有限公司 先进计算与大数据实验室, 江苏 南京 210003
  • 收稿日期:2018-04-12 修回日期:2018-06-03 出版日期:2019-01-05 发布日期:2019-01-14
  • 作者简介:李泊(1988-),女,讲师,从事计算机视觉与应用研究,E-mail:libo@njau.edu.cn
  • 基金资助:
    国家电网公司总部科技项目(基于人工智能的视频图像处理及在巡检中的应用研究,5455HJ170002)。

Broken Strands Detection of Transmission Line for Rural Distribution Network Based on Aerial Image

LI Bo1, CHEN Cheng2   

  1. 1. College of Engineering, Nanjing Agricultural University, Nanjing 210031, China;
    2. Advanced Computing and Big Data Laboratory, Global Energy Interconnection Research Institute Co. Ltd., Nanjing 210003, China
  • Received:2018-04-12 Revised:2018-06-03 Online:2019-01-05 Published:2019-01-14
  • Supported by:
    This work is supported by Science and Technology Project of SGCC(Research on Artificial Intelligence in Video Image Processing and Its Application of Power Patrol, No.5455HJ170002).

摘要: 农村配电网电力巡检所涉及的区域广阔,利用飞行平台实现基于航拍图像的电力巡检能有效提高效率,节省开支。飞行平台的拍摄角度、远近程度、光照等外部条件均不可控,给导线的断股检测带来困难。提出一种导线断股检测方法,仅通过导线轮廓信息实现断股检测,确保该方法的通用性。该方法通过LSD(line segment detector)检测图像中的导线线段,并利用线段长度来过滤非导线线段,控制了导线检测所需时间。在此基础上利用形变物体检测算法(active basis model, ABM),有效地提升了断股检测的准确率。实验结果表明所提出的方法适用于不同复杂背景和拍摄环境下的导线断股检测,在电力线路的检测中具有良好应用前景。

关键词: 农村配电网, 断股检测, 电力巡检, 航拍图像, 形变物体, 导线

Abstract: The aerial platform could significantly increase the efficiency of electric patrol and reduce its costs in rural distribution network, due to its wide area. However, the angle and distance of shooting are not controllable, and the illumination of environment is also capricious, which impose great challenges on the application of aerial electric patrol. In this paper, a contour information based universal broken strands detection method is proposed. Specifically, the segments of cable in the image are detected by LSD (line segment detector), and the length of the line segment is used to filter the non-wire line segments, controlling the total detection time. Meanwhile, the detection accuracy has been significantly improved by adopting the ABM (active basis model) algorithm. Simulation results validate the effectiveness and the practicality of the proposed methods.

Key words: rural distribution network, broken strands detection, power patrol, aerial image, deformation object, cable

中图分类号: