|
Title |
Development of a Bridge Deflection Measurement System Based on Action Cameras, Laser Beams, and Deep Learning-Based Distortion Correction
|
|
Authors |
이규완(Kyu-Wan Lee) ; 김도균(Do-Kyun Kim) ; 박영식(Young-Sik Park) |
|
DOI |
https://doi.org/10.11112/jksmi.2026.30.1.9 |
|
Keywords |
교량 처짐; 딥러닝; 상호 상관; 액션캠; 비접촉 변위 계측 Bridge deflection; Deep learning; Cross-correlation; Action camera; Non-contact displacement measurement |
|
Abstract |
This study proposes a non-contact bridge deflection measurement system utilizing a laser beam and a high-speed action camera to overcome the installation limitations of conventional contact sensors and the dynamic measurement constraints of webcam-based systems. The proposed system is configured with a 4K resolution action camera capable of 100 fps to precisely capture dynamic behaviors during high-speed traffic. Furthermore, a deep learning-based cross-correlation coordinate correction technique was applied to significantly reduce lens distortion and geometric errors. Indoor verification tests demonstrated high precision, recording an error range of approximately 0.03 to 0.04 mm in micro-displacement scenarios of less than 1 mm, which approaches the limit of hardware resolution. In field loading tests conducted on an in-service PSC I-girder bridge, the system confirmed its applicability and reliability, showing a maximum error of less than 0.04 mm under static loads and 0.05 mm under dynamic driving conditions compared to reference LVDTs. The developed system offers a cost-effective and high-precision solution for both static and dynamic displacement measurement, making it highly effective for structural health monitoring of small and medium-sized bridges.
|