Non-contact fatigue crack detection in civil infrastructure through image overlapping and crack breathing sensing

Abstract Fatigue cracks are of critical structural safety concern in civil infrastructure. Many existing fatigue crack sensing methods are contact-based, hence extensive human operation is necessary for sensor and/or actuator deployment. In this study, we propose a vision-based non-contact approach to detect fatigue cracks through image overlapping. We treat crack breathing behavior, the small cyclic movement of the crack perpendicular to the crack path under repetitive fatigue loads, as a robust indicator for crack identification. The differential image features provoked by a breathing crack can be extracted, enhanced, and visualized through a series of image processing techniques. The performance of the proposed approach is experimentally validated through two laboratory setups including a small-scale steel compact specimen and a large-scale bridge to cross-frame connection specimen. Test results demonstrate the capability of the proposed approach in reliably identifying the fatigue crack, even the true crack is surrounded by other non-crack features.