
浏览全部资源
扫码关注微信
武汉大学测绘遥感信息工程全国重点实验室,湖北 武汉,430079
Revised:2026-08-28,
Published:05 October 2026
移动端阅览
唐炉亮,叶昱彤,施泓羽,等.基于智能体的无人机视频车道级交通流探测[J].测绘地理信息,2026,51(5):9-15
TANG Luliang,YE Yutong,SHI Hongyu,et al.Agent-Based Lane-Level Traffic Flow Detection by Drone Video [J].Journal of Geomatics,2026,51(5):9-15
唐炉亮,叶昱彤,施泓羽,等.基于智能体的无人机视频车道级交通流探测[J].测绘地理信息,2026,51(5):9-15 DOI: 10.14188/j.2095-6045.20260440.
TANG Luliang,YE Yutong,SHI Hongyu,et al.Agent-Based Lane-Level Traffic Flow Detection by Drone Video [J].Journal of Geomatics,2026,51(5):9-15 DOI: 10.14188/j.2095-6045.20260440.
车道级交通流探测是交通信号控制与道路运行评价的重要基础。针对无人机视频交通流获取依赖人工检测线、相机标定和预设道路拓扑的问题,本文提出一种基于智能体的车道级交通流探测方法。该方法依据道路分支关系划分交汇与非交汇场景,利用视觉智能体定位交汇区域及边界;融合道路分割与车辆运动方向构建方向车道单元,依据轨迹访问次序生成交通流图,实现节点流量与转向流量计算。CAROM Air数据集4个场景实验表明,10个交通方向计数准确率达98.6%,非交汇路段节点流量和中型交叉口边流量相关系数分别为0.996 9和0.988 6,验证了方法的有效性。实现了由视频观测向结构化车道级交通流的自动转换。
Lane-level traffic flow detection provides an important basis for traffic signal control and road operational evaluation. To address the reliance of UAV video-based traffic flow acquisition on manually defined detection lines, camera calibration, and predefined road topology, this study proposes an agent-based method for lane-level traffic flow detection. The method distinguishes junction and non-junction scenarios according to road-branch relationships and employs a visual agent to locate junction regions and determine their boundaries. Road segmentation results are then integrated with vehicle motion directions to construct directional lane units, and a traffic flow graph is generated based on the sequential visits of vehicle trajectories to these units, enabling the computation of node flows and turning flows. Experiments on four scenarios from the CAROM Air dataset show that the counting accuracy across 10 traffic directions reaches 98.6%, while the correlation coefficients for node flows on non-merging/splitting road segments and edge flows at a medium-sized intersection reach 0.996 9 and 0.988 6, respectively, demonstrating the effectiveness of the proposed method. The method enables an automatic transformation from video observations to structured lane-level traffic flows.
Klein L A , Mills M K , Gibson D R P . Traffic Detector Handbook: Third Edition—Volume I:FHWA-HRT-06-108 [R]. McLean, VA, USA : Federal Highway Administration , 2006 , https://www.fhwa.dot.gov/publications/research/operations/its/06108/
Daganzo C F . The Cell Transmission Model, Part II: Network Traffic [J]. Transportation Research Part B: Methodological , 1995 , 29 ( 2 ): 79 - 93
黄玲 , 吴泽荣 , 洪佩鑫 , 等 . 基于地空信息融合的无人机交通状态感知方法研究 [J]. 中国公路学报 , 2021 , 34 ( 12 ): 249 - 261
Liu Fei , Zeng Zhiyuan , Jiang Rong . A Video-Based Real-Time Adaptive Vehicle-Counting System for Urban Roads [J]. PLoS One , 2017 , 12 ( 11 ): e0186098
戴喆 , 吴宇轩 , 董是 , 等 . 雷达与视觉传感器融合的高速公路全域车辆轨迹与交通参数检测方法 [J]. 交通运输工程学报 , 2025 , 25 ( 1 ): 197 - 210
Li Mengzhang , Zhu Zhanxing . Spatial-Temporal Fusion Graph Neural Networks for Traffic Flow Forecasting [J]. Proceedings of the AAAI Conference on Artificial Intelligence , 2021 , 35 ( 5 ): 4 189 - 4 196
冯宁 , 郭晟楠 , 宋超 , 等 . 面向交通流量预测的多组件时空图卷积网络 [J]. 软件学报 , 2019 , 30 ( 3 ): 759 - 769
Liu Shipeng , Wang Xingjian . An Improved Transformer Based Traffic Flow Prediction Model [J]. Scientific Reports , 2025 , 15 : 8284
Shen Guojiang , Yu Kaifeng , Zhang Meiyu , et al . ST-AFN: A Spatial-Temporal Attention Based Fusion Network for Lane-Level Traffic Flow Prediction [J]. PeerJ Computer Science , 2021 , 7 : e470
Butilă E V , Boboc R G . Urban Traffic Monitoring and Analysis Using Unmanned Aerial Vehicles (UAVs): A Systematic Literature Review [J]. Remote Sensing , 2022 , 14 ( 3 ): 620
马小龙 , 余强 , 刘建蓓 , 等 . 基于无人机视频拍摄的高速公路小型车换道行为特性 [J]. 中国公路学报 , 2020 , 33 ( 6 ): 95 - 105
Ravi N , Gabeur V , Hu Y T , et al . SAM 2: Segment Anything in Images and Videos [C]// International Conference on Learning Representations . Singapore, Singapore : International Conference on Learning Representations , 2025
Maquiling V , Byrne S A , Niehorster D C , et al . Zero-Shot Segmentation of Eye Features Using the Segment Anything Model (SAM) [J]. Proceedings of the ACM on Computer Graphics and Interactive Techniques , 2024 , 7 ( 2 ): 1 - 16
Sultan R I , Li Chengyin , Zhu Hui , et al . GeoSAM: Fine-Tuning SAM with Multi-modal Prompts for Mobility Infrastructure Segmentation [C]// ECAI 2025: 28th European Conference on Artificial Intelligence , Amsterdam, Netherlands : iOS Press , 2025
Feng Wenqing , Guan Fangli , Sun Chenhao , et al . Road-SAM: Adapting the Segment Anything Model to Road Extraction from Large Very-High-Resolution Optical Remote Sensing Images [J]. IEEE Geoscience and Remote Sensing Letters , 2024 , 21 : 6012605
戴激光 , 王杨 , 杜阳 , 等 . 光学遥感影像道路提取的方法综述 [J]. 遥感学报 , 2020 , 24 ( 7 ): 804 - 823
王国志 , 黄玉春 . 高分辨率遥感图像的道路自动提取 [J]. 测绘地理信息 , 2020 , 45 ( 3 ): 34 - 38
郑莉 , 孙梦 , 季铮 , 等 . 基于车载激光点云的自适应阈值提取道路标记方法 [J]. 测绘地理信息 , 2019 , 44 ( 4 ): 111 - 115
Qi Heng , Shi Hongyu , Yang Xue , et al . AutoLRG: A Two-Stage Framework for Automated Lane-Level Road Graph Construction [J]. IEEE Transactions on Geoscience and Remote Sensing , 2025 , 63 : 4413013
Yao S , Zhao J , Yu D , et al . ReAct: Synergizing Reasoning and Acting in Language Models [C]// International Conference on Learning Representations . Kigali, Rwanda : OpenReview.net , 2023
Chowa S S , Alvi R , Rahman S S , et al . From Language to Action: A Review of Large Language Models as Autonomous Agents and Tool Users [J]. Artificial Intelligence Review , 2026 , 59 ( 2 ): 71
Zhang Yunkai , Li Linda , Cui Yingxin , et al . Grid2Matrix: Revealing Digital Agnosia in Vision-Language Models [EB/OL]. 2026 : arXiv : 2604 . 09687 . https://arxiv.org/abs/2604.09687 https://arxiv.org/abs/2604.09687
Carion N , Gustafson L , Hu Yuanting , et al . SAM 3: Segment Anything with Concepts [EB/OL]. 2025 : arXiv : 2511 . 16719 . https://arxiv.org/abs/2511.16719 https://arxiv.org/abs/2511.16719
Lu D , Eaton E , Weg M , et al . CAROM Air—Vehicle Localization and Traffic Scene Reconstruction from Aerial Videos [C]// 2023 IEEE International Conference on Robotics and Automation (ICRA) . London, United Kingdom . New York : IEEE Press , 2023
Lu D , Jammula V C , Como S , et al . CAROM — Vehicle Localization and Traffic Scene Reconstruction from Monocular Cameras on Road Infrastructures [J]. IEEE International Conference on Robotics and Automation , 2021 : 233004562 , doi: 10.48550/arXiv.2104.00893 http://dx.doi.org/10.48550/arXiv.2104.00893
Barmpounakis E , Geroliminis N . On the New era of Urban Traffic Monitoring with Massive Drone Data: The pNEUMA Large-Scale Field Experiment [J]. Transportation Research Part C: Emerging Technologies , 2020 , 111 : 50 - 71 , doi: 10.1016/j.trc.2019.11.023 http://dx.doi.org/10.1016/j.trc.2019.11.023
Bai Xiangdong , Feng Xuyu , Yin Yuanyuan , et al . Combining Images and Trajectories Data to Automatically Generate Road Networks [J]. Remote Sensing , 2023 , 15 ( 13 ): 3343 , doi: 10.3390/rs15133343 http://dx.doi.org/10.3390/rs15133343
0
Views
4
下载量
0
CSCD
Publicity Resources
Related Articles
Related Author
Related Institution
京公网安备11010802024621