Scholarlink

<p>Aiming at the issue that existing target detection algorithms for identifying pin defects in transmission line images captured by Unmanned Aerial Vehicle (UAV) patrol flights are susceptible to omission and misdetection due to factors such as complex backgrounds and a large number of similar parts, a framework for detecting pin defects in transmission lines based on image matching is proposed. Firstly, the SpuerGlue image matching algorithm is employed to match the captured image with the reference image, while the depth estimation algorithm MiDaS is utilized to predict the depth information of the image to be detected and the reference map. This depth information is then leveraged in the deskewing process to mitigate the discrepancies arising from varying shooting angles. Subsequently, the map of the part to be detected is obtained, and ResNet50 is applied to classify it, thereby ascertaining the presence or absence of defects. Experimental results show that the method can accurately identify and categorize pin defects.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A framework for detecting defects in transmission line pins based on image matching

  • 期刊论文

  • 发表日期：

    2025年8月18日

查看全文
  • YanPeng Ji,
  • JianLi Zhao,
  • LiangShuai Liu,
  • LiBin Wang,
  • HaiYan Feng

摘要

Aiming at the issue that existing target detection algorithms for identifying pin defects in transmission line images captured by Unmanned Aerial Vehicle (UAV) patrol flights are susceptible to omission and misdetection due to factors such as complex backgrounds and a large number of similar parts, a framework for detecting pin defects in transmission lines based on image matching is proposed. Firstly, the SpuerGlue image matching algorithm is employed to match the captured image with the reference image, while the depth estimation algorithm MiDaS is utilized to predict the depth information of the image to be detected and the reference map. This depth information is then leveraged in the deskewing process to mitigate the discrepancies arising from varying shooting angles. Subsequently, the map of the part to be detected is obtained, and ResNet50 is applied to classify it, thereby ascertaining the presence or absence of defects. Experimental results show that the method can accurately identify and categorize pin defects.