Image-based bolt self-localization and bolt-loosening detection using deep learning and an improved homography-based prospective rectification method
Advances in Structural Engineering · 2023-05 · 26 (7) · 1242-1259
DOI: 10.1177/13694332231157260 ↗Kurz gefasst
Fachartikel in der Zeitschrift Advances in Structural Engineering, online erschienen am 20.02.2023. Verfasst von ChengQian Xie und 3 weiteren Autor:innen. Die Arbeit wurde bisher 11-mal zitiert (Crossref) und nennt 22 Literaturangaben. Gefördert von: Master’s Innovation Program of Chongqing University of Science and Technology, Natural Science Foundation of China, Foundation of Chongqing Science and Technology Commission.
Abstract (Originalsprache)
The bolt loosening detection method has been paid attention by engineering and academic scholars. The presented methods only focus on the identification of the bolt based on deep learning, and the problem of bolt localization and the influence of shadow on distortion correction is seldom studied. In this paper, a bolt self-localization method based on YOLOv4 deep learning algorithm is proposed. A bolt numbering rule is established and the deep learning is introduced to identify the number and locate the bolt. A new square bolt gasket is proposed and four corner points are used for distortion correction. In order to reduce the influence of shadows, a grayscale enhancement strategy is proposed to improve the correction stability. Finally, a laboratory flange joint is used to verify the proposed method. The results show that the bolt self-localization method is feasible and the new bolt gasket can effectively improve the stability of distortion correction and bolt loosening detection.
Förderung
- Master’s Innovation Program of Chongqing University of Science and Technology
- Natural Science Foundation of China
- Foundation of Chongqing Science and Technology Commission
- Research Foundation of Chongqing University of Science and Technology
- Science and Technology Research Program of Chongqing Municipal Education Commission
Zitieren
APA
Xie, C., Luo, J., Tang, K., & Zhong, Y. (2023). Image-based bolt self-localization and bolt-loosening detection using deep learning and an improved homography-based prospective rectification method. Advances in Structural Engineering, 26(7), 1242-1259. https://doi.org/10.1177/13694332231157260
Vancouver
Xie C, Luo J, Tang K, Zhong Y. Image-based bolt self-localization and bolt-loosening detection using deep learning and an improved homography-based prospective rectification method. Advances in Structural Engineering. 2023;26(7):1242-1259. doi:10.1177/13694332231157260
BibTeX
@article{Xie2023,
author = {Xie, ChengQian and Luo, Jun and Tang, KaiSen and Zhong, Yongli},
title = {Image-based bolt self-localization and bolt-loosening detection using deep learning and an improved homography-based prospective rectification method},
journal = {Advances in Structural Engineering},
year = {2023},
volume = {26},
number = {7},
pages = {1242--1259},
doi = {10.1177/13694332231157260}
}
EUVIMED