EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Regional feature analysis for automated welding defect classification: a multi-domain feature extraction approach

Mingming Zhang, Jan Polzer, Shi Cheng, Qunfeng Liu, Xun Xu

International Journal of Machine Learning and Cybernetics · 2026-11 · 17 (11)

DOI: 10.1007/s13042-026-03347-3 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Machine Learning and Cybernetics, online erschienen am 05.11.2026. Verfasst von Mingming Zhang und 4 weiteren Autor:innen. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: University of Auckland.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Automated welding defect detection is a critical challenge in manufacturing, where traditional image analysis fails to capture the spatially-dependent nature of defects. This study addresses this limitation by proposing and systematically evaluating a toolkit of five regional feature extraction methodologies. The proposed methods capture defect signatures across distinct regions. Two of them provide rich spatial descriptions: Cumulative Projection Profiling (CPP) for high-dimensional spatial analysis (6184 features) and Multi-Scale Spatial Grid Feature Extraction (MSSGFE) for hierarchical statistics (336 features). The remaining three produce compact representations: Regional Edge Direction Analysis (REDA) for edge orientations (64 features), FFT Grid Feature Extraction (FFTGFE) for frequency-domain patterns (64 features), and LBP Directional Strip Analysis (LBPDSA) for textural characteristics (32 features). A standardized preprocessing pipeline and a consistent deep learning framework were used to ensure a rigorous comparative evaluation. Experimental results show a clear performance trade-off between computational efficiency and memory footprint. The CPP, MSSGFE, and FFTGFE methods demonstrated suitability for high-throughput systems, achieving classification accuracies above 99% at processing speeds exceeding 95 FPS. In contrast, the REDA and LBPDSA methods, while slower, produced extremely compact feature vectors (

Förderung

  • University of Auckland

Zitieren

APA

Zhang, M., Polzer, J., Cheng, S., Liu, Q., & Xu, X. (2026). Regional feature analysis for automated welding defect classification: a multi-domain feature extraction approach. International Journal of Machine Learning and Cybernetics, 17(11). https://doi.org/10.1007/s13042-026-03347-3

Vancouver

Zhang M, Polzer J, Cheng S, Liu Q, Xu X. Regional feature analysis for automated welding defect classification: a multi-domain feature extraction approach. International Journal of Machine Learning and Cybernetics. 2026;17(11). doi:10.1007/s13042-026-03347-3

BibTeX
@article{Zhang2026,
  author = {Zhang, Mingming and Polzer, Jan and Cheng, Shi and Liu, Qunfeng and Xu, Xun},
  title = {Regional feature analysis for automated welding defect classification: a multi-domain feature extraction approach},
  journal = {International Journal of Machine Learning and Cybernetics},
  year = {2026},
  volume = {17},
  number = {11},
  doi = {10.1007/s13042-026-03347-3}
}
Kostenlos starten →