| 13 | 0 | 38 |
| 下载次数 | 被引频次 | 阅读次数 |
车道线的准确检测是智能交通系统与自动驾驶环境感知的关键。针对当前纯视觉检测方法在城市复杂环境因遮挡、光照变化等导致准确度与鲁棒性不足的问题，本文提出一种基于IAFNet+LSTM的多模态融合车道线检测方法。通过设计轻量级图像—角度交叉注意力模块(LIA-CAM)实现视觉特征与车辆转向角信息的跨模态深度融合，并引入LSTM对转向角序列进行时序建模，以有效捕捉车辆动态意图。基于ImageAngle-Udacity数据集的6 308张道路图像，在强光、弱光、急转弯及车辆遮挡等多种复杂场景下，将本文方法与IAFNet、LaneNet、SCNN等方法进行对比试验和参数敏感性分析。结果显示，本文方法的IoU为0.788,精确率为0.878,召回率为0.896,F1值为0.867,相比最优基线模型IAFNet分别提升约1.6%、2.1%、3.9%、1.6%;在复杂光照和几何条件下具有良好鲁棒性，相比传统单模态方法具有显著优势，可为自动驾驶系统提供可靠的车道线感知能力。
Abstract：Accurate lane detection is a fundamental component of intelligent transportation systems and autonomous driving environment perception.To address the degraded accuracy and insufficient robustness of conventional vision-only methods caused by object occlusions and drastic illumination changes in complex urban environments, this paper proposes a multi-modal fusion lane detection approach based on IAFNet and LSTM.The proposed method introduces a lightweight image-angle cross-attention module(LIA-CAM) to achieve deep cross-modal integration of visual features and vehicle steering angle information, and employs LSTM to model the temporal sequence of steering angles, thereby effectively capturing the vehicle′s dynamic intention.Using 6 308 road images from the ImageAngle-Udacity dataset, comparative experiments and parameter sensitivity analysis are conducted under various complex scenarios, including high glare, low light, sharp turns, and vehicle occlusions.The proposed method is evaluated against mainstream methods such as IAFNet, LaneNet, and SCNN.Experimental results show that the proposed method achieves an IoU of 0.788,precision of 0.878,recall of 0.896,and F1-score of 0.867.Compared to the best-performing baseline IAFNet, the proposed method improves the above metrics by approximately 1.6%,2.1%,3.9%,and 1.6%,respectively.These results demonstrate that the proposed method exhibits strong robustness under complex lighting and geometric conditions, offering significant advantages over traditional single-modal approaches and providing reliable lane perception capability for autonomous driving systems.
[1] 陈婕，刘纪平，徐胜华.增强边缘信息的全卷积神经网络遥感影像建筑物变化检测[J].测绘通报，2023(6):61-67.
[2] 王庆，曾琪明，张海真，等.全极化SAR图像边缘检测的随机距离法[J].测绘学报，2015,44(7):753-760.
[3] Ozgunalp U.Combination of the symmetrical local threshold and the sobel edge detector for lane feature extraction[C]//2017 9th International Conference on Computational Intelligence and Communication Networks (CICN).Girne:IEEE,2017:24-28.
[4] Low C Y,Zamzuri H,Mazlan S A.Simple robust road lane detection algorithm[C]//2014 5th International Conference on Intelligent and Advanced Systems (ICIAS).Kuala Lumpur:IEEE,2014:1-4.
[5] Hur J,Kang S N,Seo S W.Multi-lane detection in urban driving environments using conditional random fields[C]//2013 IEEE Intelligent Vehicles Symposium (IV).Gold Coast:IEEE,2013:1297-1302.
[6] Fan Chao,Song Yupei,Jiao Yajie.Multi-lane detection based on deep convolutional neural network[J].IEEE Access,2019,7:150833-150841.
[7] 石金鹏，张旭.基于空间语义分割的多车道线检测跟踪网络[J].光学精密工程，2023,31(9):1357-1365.
[8] Neven D,De Brabandere B,Georgoulis S,et al.Towards end-to-end lane detection:an instance segmentation approach[C]//2018 IEEE Intelligent Vehicles Symposium (IV).Changshu:IEEE,2018:286-291.
[9] Yoo S,Lee H S,Myeong H,et al.End-to-end lane marker detection via row-wise classification[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).Seattle:IEEE,2020:4335-4343.
[10] Tabelini L,Berriel R,Paixão T M,et al.Keep your eyes on the lane:real-time attention-guided lane detection[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Nashville:IEEE,2021:294-302.
[11] Liu Ruijin,Yuan Zejian,Liu Tie,et al.End-to-end lane shape prediction with transformers[C]//2021 IEEE Winter Conference on Applications of Computer Vision (WACV).Waikoloa:IEEE,2021:3693-3701.
[12] Baek S W,Kim M J,Suddamalla U,et al.Real-time lane detection based on deep learning[J].Journal of Electrical Engineering & Technology,2022,17(1):655-664.
[13] Qin Zequn,Wang Huanyu,Li Xi.Ultra fast structure-aware deep lane detection[C]//Computer Vision-ECCV 2020.Cham:Springer,2020:276-291.
[14] 陈广秋，刘枫铭，段锦，等.基于轻量化Transformer的车道线检测方法[J].华中科技大学学报(自然科学版),2025,53(3):117-126.
[15] Dong Yongqi,Patil S,Van Arem B,et al.A hybrid spatial-temporal deep learning architecture for lane detection[J].Computer-Aided Civil and Infrastructure Engineering,2023,38(1):67-86.
[16] Kim J,Park C.End-to-end ego lane estimation based on sequential transfer learning for self-driving cars[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).Honolulu:IEEE,2017:1194-1202.
[17] Hou Yuenan,Ma Zheng,Liu Chunxiao,et al.Learning lightweight lane detection CNNs by self attention distillation[C]//2019 IEEE/CVF International Conference on Computer Vision (ICCV).Seoul:IEEE,2019:1013-1021.
[18] 方遒，李伟林，梁卓凡，等.基于多尺度复合卷积和图像分割融合的车道线检测算法[J].北京理工大学学报，2023,43(8):792-802.
[19] 武小兰，孙奔奔，白志峰，等.基于多尺度特征融合和组合结构损失优化的深度学习弯道车道线检测方法研究[J].控制与决策，2025,40(8):2467-2472.
[20] 潘玉恒，李慧洁，鲁维佳，等.基于级联残差的弯道增强车道线检测[J].测绘科学，2024,49(7):37-44.
[21] Pan Xingang,Shi Jianping,Luo Ping,et al.Spatial as deep:spatial CNN for traffic scene understanding[C]//32nd AAAI Conference on Artificial Intelligence.Palo Alto:AAAI Press,2017.
[22] Garnett N,Cohen R,Pe′er T,et al.3D LaneNet:end-to-end 3D multiple lane detection[C]//2019 IEEE/CVF International Conference on Computer Vision (ICCV).Seoul:IEEE,2019:2921-2930.
[23] 刘宏建，张伟斌.基于改进LaneNet网络的车道线检测方法[J].现代交通与冶金材料，2025,5(2):39-50.
[24] Chen Zhenpeng,Liu Qianfei,Lian Chenfan.PointLaneNet:efficient end-to-end CNNs for accurate real-time lane detection[C]//2019 IEEE Intelligent Vehicles Symposium (IV).Paris:IEEE,2019:2563-2568.
[25] 李超，唐雅琪，彭琴.复杂路况下的车道线高精度检测方法[J].计算机工程与应用，2025,61(15):269-277.
[26] Wang Ze,Ren Weiqiang,Qiu Qiang.Lanenet:real-time lane detection networks for autonomous driving[EB/OL].(2018-07-04)[2025-04-24].https://arxiv.org/abs/1807.01726.
[27] Chen Long,Zhang Hanwang,Xiao Jun,et al.SCA-CNN:spatial and channel-wise attention in convolutional networks for image captioning[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Honolulu:IEEE,2017:6298-6306.
[28] Zheng Tu,Fang Hao,Zhang Yi,et al.RESA:recurrent feature-shift aggregator for lane detection[C]//Thirty-Fifth AAAI Conference on Artificial Intelligence.Palo Alto:AAAI Press,2021:3547-3554.
[29] Munir F,Azam S,Jeon M,et al.LDNet:end-to-end lane marking detection approach using a dynamic vision sensor[J].IEEE Transactions on Intelligent Transportation Systems,2022,23(7):9318-9334.
[30] Zhou Fengyun,Deng Honggui,Xu Qiguo,et al.CNTR-YOLO:improved YOLOv5 based on ConvNext and transformer for aircraft detection in remote sensing images[J].Electronics,2023,12(12):2671.DOI:10.3390/electronics12122671.
[31] He Haoyu,Cai Jianfei,Pan Zizheng,et al.Dynamic focus-aware positional queries for semantic segmentation[C]//2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Vancouver:IEEE,2023:11299-11308.
基本信息:
中图分类号:U463.6;TP391.41
引用信息:
[1]龙彩霞,张用川,郑豆豆,等.面向复杂场景融合视觉与转向角时序信息的车道线检测方法[J].地理与地理信息科学,2026,42(04):17-26.
基金信息:
重庆市规划和自然资源局科研项目“基于多场景驱动的规划自然资源高质量数据集建设关键技术研究”(KJ-2025016);重庆市规划和自然资源局科研项目“空地协同的山地城市高精地图智能构建技术研究”(KJ-2026017)
2026-07-25
2026-07-25