Confidence-feedback-weighted graph matching network: online-offline laser-induced damage site matching under complex interference

2026-06-28

AI Summary

This work addresses the challenge of accurately matching ground-truth damage locations in offline analysis of terminal optical element inspection images from high-power laser systems, where numerous pseudo-damage artifacts closely resemble real defects. To tackle this issue, the authors propose a confidence-feedback weighted graph matching network that operates solely on damage centroid coordinates. The method iteratively estimates node matching confidence and feeds it back as reliability weights during edge feature aggregation, thereby effectively suppressing interference propagation from spurious points. Geometric consistency constraints are integrated to rectify falsely confident matches, while a hard negative mining loss enhances discrimination among structurally similar damage instances. Evaluated on the Complex-Scene dataset, the approach achieves a matching F1 score of 96.36%, demonstrating significantly improved robustness and efficiency.

Abstract

Online inspection images of final optics in high-power laser facilities contain pseudo-damage sites that closely resemble true damage sites. Determining the authenticity of online-detected sites is therefore difficult and requires accurate matching to offline ground-truth sites. However, this matching remains highly challenging due to limited match-discriminative features, local geometric distortions, and numerous distractor sites. Existing matching models mainly suppress distractors implicitly through loss-function supervision. We propose a confidence-feedback-weighted graph matching network that requires only damage-site centroid coordinates as input. It estimates node matchability confidence from each round of matching scores and feeds it back as a reliability weight to guide subsequent edge-feature aggregation, thereby suppressing distractor propagation and enhancing cross-graph discriminability. Within this framework, a geometric consistency constraint calibrates spurious high-confidence matchability estimates, while a hard-example mining loss improves discrimination between structurally similar sites. Experiments on our Complex-Scene dataset show that the proposed method achieves a matching F1-score of 96.36$\%$ with robust and efficient performance.

Problem

• graph matching
• laser-induced damage
• online-offline matching
• distractor suppression
• geometric distortion

Innovation

• graph matching network
• confidence feedback
• laser-induced damage detection
• geometric consistency constraint
• hard-example mining