A Boundary-Guided Feature Modulation Network for Weld Radiographic Defect Segmentation
Applied Sciences · 01.07.2026 · 16 (13) · 6579
DOI: 10.3390/app16136579 ↗Kurz gefasst
Fachartikel in der Zeitschrift Applied Sciences, online erschienen am 01.07.2026. Verfasst von Xuanyu Yang und 6 weiteren Autor:innen. Die Arbeit nennt 26 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Accurate pixel-level segmentation of weld defects in radiographic images is essential for automated non-destructive testing (NDT) and quantitative weld-quality assessment. However, this task remains challenging because weld defects often exhibit severe foreground–background imbalance, ambiguous boundaries, weak grayscale contrast, and defect-like weld background structures, which can lead to boundary over-expansion and false-positive predictions. To address these issues, this paper proposes a boundary-guided feature modulation framework with false-positive suppression for weld radiographic defect segmentation. The method constructs boundary bands from training annotations and uses them only as label-derived training-time regularizers for attention-driven feature modulation and region-aware optimization; during validation and testing, no ground-truth mask, boundary band, or predicted boundary map is provided to the model. Multi-scale feature fusion is used to recover weak defect responses, boundary-guided dual attention enhances boundary-sensitive feature representation, and a false-positive suppression loss penalizes foreground leakage above a tolerated confidence margin in stable non-boundary background regions. Experiments on a real-world pipeline weld radiographic dataset containing 10,079 images show that the proposed method achieves a Dice score of 0.810±0.003, a Precision of 0.809±0.004, and a Surface Dice at 3 pixels of 0.394±0.008, outperforming representative CNN-based and Transformer-based segmentation baselines. Ablation studies, qualitative visualization, and distance-based false-positive analysis further demonstrate that the proposed framework improves contour reliability and reduces background false positives.
Zitieren
APA
Yang, X., Yang, F., Wu, J., Rong, R., Hu, W., Shen, Y., & Hu, J. (2026). A Boundary-Guided Feature Modulation Network for Weld Radiographic Defect Segmentation. Applied Sciences, 16(13), 6579. https://doi.org/10.3390/app16136579
Vancouver
Yang X, Yang F, Wu J, Rong R, Hu W, Shen Y, et al. A Boundary-Guided Feature Modulation Network for Weld Radiographic Defect Segmentation. Applied Sciences. 2026;16(13):6579. doi:10.3390/app16136579
BibTeX
@article{Yang2026,
author = {Yang, Xuanyu and Yang, Fan and Wu, Junjie and Rong, Rong and Hu, Wang and Shen, Yuncheng and Hu, Junjie},
title = {A Boundary-Guided Feature Modulation Network for Weld Radiographic Defect Segmentation},
journal = {Applied Sciences},
year = {2026},
volume = {16},
number = {13},
pages = {6579},
doi = {10.3390/app16136579}
}
EUVIMED