3D Measurement System for Tunnel Segment Joints Based on Enhanced Randomized Hough Transform

Monitoring differential settlement in immersed tunnel segment joints is critical for structural safety. Traditional manual inspection has low efficiency and high costs. Existing mobile automated monitoring cannot achieve submillimeter accuracy. In this study, this technical difficulty is solved and we propose a novel 3D co-ordinate measurement system. It combines improved random Hough transform with binocular vision. We developed a circularity-based ROI constraint model. An octant-interval random sampling strategy was implemented. This improves circle detection efficiency to 380.54 ms per frame. The method reduces parameter estimation errors in complex environments. Experiments show high static measurement accuracy. Errors are ±0.09 mm (X-axis), ±0.08 mm (Y-axis), and ±0.27 mm (Z-axis). Dynamic measurements maintain errors below 0.3 mm across full tunnel sections. The system was successfully tested in Dalian Bay Undersea Tunnel. It enables intelligent monitoring of joint differential settlement and opening/closing displacements. The solution offers high precision and efficiency. This technology advances automated, quantitative structural health monitoring for tunnels.