A machine vision-based method for displacement monitoring under fog interference

Accurate displacement monitoring of bridges in foggy environments remains a critical challenge for structural health monitoring. This study proposes a machine vision-based technique that integrates red-light active illumination with sub-pixel algorithms to address this issue. To mitigate the low signal-to-noise ratio caused by fog scattering, the core of our approach employs a high-brightness monochromatic red LED as an active source to provide a high-contrast target against scattered ambient noise. A sophisticated sub-pixel localization pipeline incorporating dynamic threshold searching, morphological denoising, and multi-algorithm fusion achieves a remarkable displacement resolution of 0.1 mm. Experimental results demonstrate the robustness of method across various fog conditions (light, moderate, and thick fog) with maximum RMSE below 0.0262 mm. Validation on a scaled cable-stayed bridge model shows strong agreement with laser displacement sensor measurements under dynamic loading. This research provides a cost-effective, reliable solution for bridge displacement monitoring in adverse weather conditions, advancing machine vision applications in infrastructure monitoring.