Automated underwater concrete crack width measurement system based on dual lasers and DeepCrack network

To address the challenges of low automation in underwater dam crack measurement and errors caused by multi-interface refraction distortion, this study proposes an automated measurement method based on “imaging calibration - intelligent segmentation - precise measurement.” First, an underwater light propagation model was developed using Snell’s law, which enables the transformation between pixel coordinates and real-world coordinates. By integrating underwater camera calibration and refraction compensation, refraction-induced distortion is effectively minimized. Next, a high-precision crack segmentation model, combined with dual laser technology, enables automated crack width measurement. Comparative experiments on real datasets validate the superior performance of DeepCrack in underwater crack segmentation, and underwater optical darkroom measurement experiments show that the proposed method achieves millimeter-level accuracy within a measurement range of 0.5 to 2.0 m. Additionally, the developed hardware and software for underwater crack measurement were successfully applied in real scenarios, providing reliable technical support for visualizing underwater crack measurement.