A Method of Insulator Detection from Aerial Images

Yongjie Zhai, Wu Yang, Hongkai Chen, Xiaoguang ZhaoSHILAP 2014

Abstract

With increased image processing and sensor technologies, it is possible for photogrammetrists to monitor the status of electrical power infrastructures. Identifying and extracting insulators from complex background have great potential in automatic power grid surveillance on board unmanned aerial vehicles. This paper proposes an improved method to automatically extract the insulators from aerial images. A 3D model is introduced into the existing algorithms to obtain samples of insulators, which are Haar features and AdaBoost. Finally, in order to evaluate the performances, we compare our 3D model-based method with the existing methods that does not use 3D model to show the performance improvement. The experimental results demonstrate that the proposed method is able to successfully extract the insulators from aerial images with an increased average detection ratio to 83 %.
BibTeX
@inproceedings{zhai2014method,
  author = {Yongjie Zhai and Wu Yang and Hongkai Chen and Xiaoguang Zhao},
  title = {{A Method of Insulator Detection from Aerial Images}},
  booktitle = {SHILAP},
  year = {2014},
}

Loading related content...

Figure