东南大学学报（自然科学版）2025，Vol.55Issue(5)：1311-1318,8.DOI:10.3969/j.issn.1001-0505.2025.05.011
基于计算机视觉的车致桥梁挠度测试与车辆定位研究
Measurement of vehicle-induced bridge deflection and vehicle localization based on computer vision
摘要
Abstract
To improve the convenience and accuracy of bridge deflection measurement and vehicle localiza-tion,a computer vision-based method for vehicle-induced bridge deflection measurement and vehicle localiza-tion was proposed.An unmanned aerial vehicle(UAV)was used to capture vehicle images,and the simulta-neous localization and mapping(SLAM)algorithm combined with the YOLO algorithm was used to estimate camera pose and achieve real-time vehicle localization.For bridge deflection measurement,an improved scale-invariant feature transform(SIFT)algorithm was proposed to extract and match bridge image features.The bridge deflection response was recognized by the movement of bridge features.The feasibility and accu-racy of this method were verified by the experiment of vehicle moving across the bridge.The results show that removing dynamic features from UAV imagery enhances the accuracy of pose estimation,reducing vehicle lo-calization errors to below 1%.The improved SIFT algorithm effectively extracts and matches features,achiev-ing image deflection measurement with an accuracy of approximately±0.368 pixels at a shooting distance of 1.5 m.The proposed computer vision-based deflection measurement and vehicle localization method com-bines non-contact measurement and high precision,offering reliable data support for rapid assessment of small and medium-span bridges.关键词
桥梁工程/移动车辆荷载/车辆定位/挠度测试/计算机视觉/无人机Key words
bridge engineering/moving vehicle load/vehicle localization/deflection measurement/com-puter vision/unmanned aerial vehicle分类
交通工程引用本文复制引用
万华平,王灿,房天乐,曹素功,王宁波..基于计算机视觉的车致桥梁挠度测试与车辆定位研究[J].东南大学学报（自然科学版）,2025,55(5):1311-1318,8.基金项目
国家自然科学基金优秀青年基金资助项目(52422804) （52422804）
公路桥隧智能运维技术浙江省工程中心开放基金资助项目(202401G). （202401G）
