A Vision-Based Detection-Localization-Measurement Framework for Quantitative Size Characterization of Welding Microdefects
IEEE Transactions on Instrumentation and Measurement · 2026 · 75 · 5013614-5013614
DOI: 10.1109/tim.2026.3709483 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Instrumentation and Measurement, 2026. Verfasst von Yindi Cai und 6 weiteren Autor:innen. Die Arbeit nennt 29 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
Accurate quantitative characterization of welding microdefects is essential to ensure the reliability of welded components. However, the small size of microdefects and the complexity of the welding surface induce challenges to vision-based measurement. A vision-based detection-localization-measurement framework is proposed for quantitative size characterization of welding microdefects. The framework integrates an improved YOLOv9 model for high-precision defect detection and localization with an edge-based geometric measurement algorithm. The improved YOLOv9 model is enhanced to improve contour representation consistency and spatial localization accuracy, which are critical for microdefect size estimation. Based on accurate defect localization, microdefect contours are extracted and characteristic dimensions are calculated. The effectiveness of the proposed method for microscale quantitative measurement under complex industrial conditions is demonstrated. The proposed method achieves relative measurement errors within 16 % and 12.4 % for porosity and slag inclusion defects, respectively. Source code is released in https://github.com/YouOnlyLookOnce-crypto/Welding-microdefects
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Cai, Y., Li, P., Niu, C., Li, S., Zhu, X., Dong, Z., & Kang, R. (2026). A Vision-Based Detection-Localization-Measurement Framework for Quantitative Size Characterization of Welding Microdefects. IEEE Transactions on Instrumentation and Measurement, 75, 5013614-5013614. https://doi.org/10.1109/tim.2026.3709483
Vancouver
Cai Y, Li P, Niu C, Li S, Zhu X, Dong Z, et al. A Vision-Based Detection-Localization-Measurement Framework for Quantitative Size Characterization of Welding Microdefects. IEEE Transactions on Instrumentation and Measurement. 2026;75:5013614-5013614. doi:10.1109/tim.2026.3709483
BibTeX
@article{Cai2026,
author = {Cai, Yindi and Li, Pengrui and Niu, Chao and Li, Shixiong and Zhu, Xionglong and Dong, Zhigang and Kang, Renke},
title = {A Vision-Based Detection-Localization-Measurement Framework for Quantitative Size Characterization of Welding Microdefects},
journal = {IEEE Transactions on Instrumentation and Measurement},
year = {2026},
volume = {75},
pages = {5013614--5013614},
doi = {10.1109/tim.2026.3709483}
}
EUVIMED