随着人工智能技术的快速发展，自动驾驶汽车正逐渐走入大众视野，交通行业也朝 着智能化和无人化的方向稳步迈进。车道线检测是实现车道预警、车道保持、自动巡航 等多种自动驾驶辅助系统的前提，其意义重大。随着研究的深入和技术要求的不断提升，对于车道线检测算法的准确性、实时性和鲁棒性问题仍然是一个值得关注的研究课题。本文以提高车道线检测的准确性、高效性和稳定性为目标，设计一种高精度、高效率、高稳定性的车道线检测跟踪算法。论文研究的主要内容如下：

（1）本文针对当前车道线检测算法中精度和速度的问题，设计改进 LSTR 车道线检测算法。首先针对特征提取网络设计 GSE-Block，将普通卷积替换成具有良好性能的改进 Ghost 卷积，在网络性能保持的基础上实现轻量化设计；设计一种新的批量归一化模块 Self-BN，通过保留更多的原始特征信息实现检测准确率的提高；引入 ECA 通道注意力机制，加强对车道线细节信息和边缘信息的提取效果，实现对车道线检测准确率的提高。其次对 LSTR 算法的 MLP 结构进行改进，优化其激活函数，并加入 Self-BN 模块，以达到加快模型收敛速度，提高训练效率的目的。

（2）本文针对当前车道线检测算法鲁棒性低，易造成误检漏检问题，设计基于改进卡尔曼滤波的车道线跟踪算法。首先，设计一种自适应更新卡尔曼增益的改进算法，提升卡尔曼滤波对车道线的跟踪性能。其次，设计基于多维卡尔曼滤波的车道线方程参数跟踪算法，实现对方程参数的实时跟踪预测，提高方程参数输出的稳定性和检测算法的准确率。最后，设计基于扩展卡尔曼滤波的车道线关键坐标点跟踪算法，实现对车道线的关键坐标点实时跟踪预测，进一步提升车道线检测算法的稳健性。

（3）使用公开数据集对本文设计算法进行实验验证。结果表明：对于改进的检测算法，与原网络相比，检测准确率提升 0.97%，达到 97.12%；FPS 提升了 26，达到 384FPS。对于车道线跟踪算法，与改进检测算法相比，提高了 0.90%的检测准确率，减少了 0.27%误检率和 0.59%的漏检率。本文设计算法，通过实验证明具有性能上的提升和更高的鲁棒性，对自动驾驶的研究具有参考意义和应用价值。

With the rapid development of artificial intelligence technology, self-driving cars are gradually coming into the public's view, and the transport industry is steadily progressing towards intelligence and unmanned. Lane line detection is the prerequisite for the realization of various automatic driving assistance systems such as lane warning, lane keeping, and automatic cruise control, which is of great significance. With the deepening of research and the continuous improvement of technical requirements, the accuracy, real-time, and robustness of lane line detection algorithms are still a research topic that deserves attention. In this thesis, we aim to improve the accuracy, efficiency, and stability of lane line detection, and design a lane line detection tracking algorithm with high accuracy, high efficiency and high stability. The main contents of the thesis research are as follows:

(1) In this thesis, for the current lane line detection algorithm in the accuracy and speed of the problem, design to improve the LSTR lane line detection algorithm. Firstly, GSE-Block is designed for the feature extraction network, replacing the ordinary convolution with the improved Ghost convolution with good performance to achieve a lightweight design on the basis of maintaining the network performance; the new batch normalization module Self-BN is designed to improve the detection accuracy by retaining more original feature information; the ECA channel attention mechanism is introduced to enhance the extraction effect of lane line The ECA channel attention mechanism is introduced to enhance the extraction effect of detail information and edge information to achieve the improvement of lane line detection accuracy. Secondly, the MLP structure of the LSTR algorithm is improved, the activation function is optimized, and the Self-BN module is added to accelerate the convergence speed of the model and improve the training efficiency.

(2) In this thesis, for the current lane line detection algorithm robustness is low, easy to cause misdetection leakage detection problem, design based on improved Kalman filter lane line tracking algorithm. Firstly, design an improved algorithm to adaptively update the Kalman gain to improve the tracking performance of Kalman filter for lane lines. Secondly, design the lane line equation parameter tracking algorithm based on multidimensional Kalman filtering to achieve real-time tracking and prediction of the equation parameters and improve the stability of the equation parameter output and the accuracy of the detection algorithm. Finally, the tracking algorithm of key coordinate points of lane lines based on extended Kalman filtering is designed to achieve real-time tracking prediction of key coordinate points of lane lines, which further improves the robustness of the lane line detection algorithm.

(3) Experimental validation of the algorithms designed in this thesis is carried out using publicly available datasets. The results show that: for the improved detection algorithm, compared with the original network, the detection accuracy is improved by 0.97% to 97.12%; the FPS is improved by 26 to 384 FPS. for the lane line tracking algorithm, compared with the improved detection algorithm, it improves the detection accuracy by 0.90%, and reduces the false positive rate by 0.27% and the false negative rate by 0.59%. The algorithm designed in this thesis is experimentally proved to have performance improvement and higher robustness, which is of reference and application value for the research of automatic driving.

[1] 中华人民共和国国家统计局. 国家数据(运输和邮电)[EB/OL]. https: //data. stats. gov. cn/ easyquery.htmcn=C01&zb=A0S0D01&sj=2023, [2024.3.14].

[2] 宁华晶. 不安全驾驶行为与道路交通事故之间的关系研究[D]. 兰州: 兰州交通大学, 2023.

[3] Torres LA, Murilo A, Lopes RV. Analysis and Virtual Validation of Vehicle Dynamics Models for Electronic Stability Control[J]. IEEE Latin America Transactions, 2024, 22(2):166-172.

[4] 童成前. 汽车 ABS 系统故障分析[J]. 汽车电器, 2020, 1(8): 78-79.

[5] Cheng Z J, Jeng L W, Li K. Behavioral Classification of Drivers for Driving Efficiency Related ADAS Using Artificial Neural Network[C]// 2018 IEEE International Conference on Advanced Manufacturing (ICAM), IEEE. 2018: 173-176.

[6] 中华人民共和国工业和信息化部, 汽车驾驶自动化分级: GB/T 40429-2021[S]. 2022.

[7] 王贵林,韩同群.智能汽车车道保持控制策略研究[J]. 汽车零部件, 2023,1(8):20-26.

[8] 张磊, 彭金栓, 宋臻. 车道变换行为研究进展[J]. 重庆理工大学学报(自然科学), 2022,36(11): 40-50.

[9] 廖柳芳. 车辆自适应巡航和主动换道控制策略的研究[D]. 秦皇岛: 燕山大学, 2023.

[10] Sapkal, Ashwini, Arti. Lane Detection Techniques for Self-driving Vehicle: Comprehensive Review[J]. Multimedia Tools and Applications, 2023, 82(22): 33983-34004.

[11] Otsuka Y, Muramatsu S, Takenaga H. Multitype Lane Markers Recognition Using Local Edge Direction[C]// 2002 Intelligent Vehicle Symposium, IEEE. 2002: 604-609.

[12] Parajuli A, Celenk M, Riley H B. Robust Lane Detection in Shadows and Low Illumination Conditions using Local Gradient Features[J]. Open Journal of Applied Sciences, 2013, 1(1):68-74.

[13] 王雨蔷. 复杂环境下的车道线检测方法研究[D]. 广州: 华南理工学, 2021.

[14] 丁蒙, 戴曙光, 曾思浩. 基于自适应 ROI 和蚁群边缘检测算法的车道线检测[J]. 软件, 2019, 40(12): 124-128.

[15] Ke H M. Research on Lane Detection and Tracking Under Complex Road Conditions[J]. Frontiers in Science and Engineering, 2021, 1(3):162-165.

[16] Kanopoulos N, Vasanthavada N, Baker R L. Design of an Image Edge Detection Filter Using the Sobel Operator[J]. IEEE Journal of Solid-State Circuits, 1988, 23(2): 358-367.

[17]Javeed MA, Ghaffar MA, Ashraf MA. Lane Line Detection and Object Scene Segmentation Using Otsu Thresholding and the Fast Hough Transform for Intelligent Vehicles in Complex Road Conditions[J]. Electronics, 2023, 12(5): 1079-1090.

[18] Canny J. A Computational Approach to Edge Detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1986, 8(6): 679-698.

[19] Illingworth, John, Josef Kittler. A Survey of the Hough Transform[J]. Computer vision, graphics, and image processing, 1988, 44(1): 87-116.

[20] 李楠, 顾苏菁, 叶飞. 一种基于 HSI 颜色模型的车道线检测算法[J]. 信息技术与信息化,2022(4): 58-61.

[21] 穆柯楠, 王会峰, 赵祥模. 成像模型约束非均匀 B 样条曲线拟合车道线检测[J]. 公路交通科技, 2022, 39(9): 132-139.

[22] Zhou, Shengyan, Yanhua Jiang. A Novel Lane Detection Based on Geometrical Model and Gabor Filter[C]// 2010 IEEE Intelligent Vehicles Symposium, IEEE. 2010: 59-64.

[23] 洪伟, 王吉通, 刘宇. 基于 DBSCAN 的复杂环境下车道线鲁棒检测及跟踪[J]. 吉林大学学报(工学版), 2020, 50(6): 2122-2130.

[24] 刘富强, 张姗姗, 朱文红. 一种基于视觉的车道线检测与跟踪算法[J]. 同济大学学报(自然科学版), 2010, 38(2): 223-229.

[25] Van Gansbeke, Wouter, Bert De Brabandere. End-to-end Lane Detection Through Differentiable Least-squares Fitting[C]// 2019 Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops, IEEE. 2019: 1-11.

[26] Wang H, Wang Y, Zhao X. Lane Detection of Curving Road for Structural Highway with Straight-Curve Model on Vision[J]. IEEE Transactions on Vehicular Technology. 2019, 68(6): 5321-5330.

[27] Chae YJ, Park SJ, Kang ES. Point2Lane: Polyline-Based Reconstruction with Principal Points for Lane Detection[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(12): 14813-14829.

[28] Huval B, Wang T, Tandon S. An Empirical Evaluation of Deep Learning on Highway Driving[J]. arXiv preprint arXiv: 1504.01716, 2015.

[29] Yoo S, Lee H S, Myeong H. End-to-end Lane Marker Detection Via Row-wise

Classification[C]// Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops, IEEE. 2020: 1006-1007.

[30] Qin Z, Wang H, Li X. Ultra Fast Structure-aware Deep Lane Detection[C]// 2020 Computer Vision–ECCV, Proceedings, IEEE. 2020: 276-291.

[31] Hou Y, Ma Z, Liu C. Inter-region Affinity Distillation for Road Marking Segmentation[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE. 2020: 12486-12495.

[32] Wang J, Ma Y, Huang S. A Key Point-based Global Association Network for Lane Detection[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE. 2022: 1392-1401.

[33] Tabelini L, Berriel R, Paixao T M. Keep Your Eyes on the Lane: Real-time Attention-guided Lane Detection[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE. 2021: 294-302.

[34] Zheng T, Huang Y, Liu Y. Clrnet: Cross Layer Refinement Network for Lane Detection[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE. 2022: 898-907.

[35] Tian Y, Gelernter J, Wang X. Lane Marking Detection Via Deep Convolutional Neural Network[J]. Neurocomputing, 2018, 280(1): 46-55.

[36] Li X, Li J, Hu X. Line-CNN: End-to-End Traffic Line Detection with Line Proposal Unit[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 21(1): 248-258.

[37]Jiang P, Ergu D, Liu F. A Review of Yolo Algorithm Developments[J]. Procedia Computer Science, 2022, 199(1): 1066-1073.

[38] Liu W, Anguelov D, Erhan D. SSD: Single Shot Multi Box Detector[C]// Computer Vision–ECCV 2016, IEEE. 2016: 21-37.

[39] Xingang Pan, Jianping Shi, Ping Luo. Spatial As Deep: Spatial CNN for Traffic Scene Understanding[J]. AAAI Conference on Artificial Intelligence, 2018, 32(1): 7276-7283.

[40] Liu L, Chen X, Zhu S, et al. CondLaneNet: a Top-to-down Lane Detection Framework Based on Conditional Convolution[C]// Proceedings of the IEEE/CVF International Conference on Computer Vision, IEEE. 2021: 3773-3782.

[41] Neven D, De Brabandere B, Georgoulis S. Towards End-to-End Lane Detection: An Instance Segmentation Approach[C]// 2018 IEEE intelligent vehicles symposium, IEEE. 2018: 286-291.

[42] Xu H, Wang S, Cai X. CurveLane-NAS: Unifying Lane-Sensitive Architecture Search and Adaptive Point Blending[C]// Computer Vision-ECCV 2020, IEEE. 2020: 689-704.

[43] Liu R, Yuan Z, Liu T, et al. End-to-end Lane Shape Prediction with Transformers[C]// Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, IEEE. 2021: 3694-3702.

[44] Tabelini L, Berriel R, Paixao T M. PolyLaneNet: Lane Estimation via Deep Polynomial Regression[C]// International Conference on Pattern Recognition. IEEE Computer Society, IEEE. 2021: 6150-6156.

[45] Yang X, Li Q, He C. Improved Meanshift Tracking Algorithm Based on Optical Flow[C]// 2023 IEEE 6th Eurasian Conference on Educational Innovation (ECEI), IEEE. 2023: 159-162.

[46] Sahl, Samer, Enbin Song. Robust Cubature Kalman Filter for Moving-Target Tracking with Missing Measurements[J]. Sensors, 2024, 24(2): 392-403.

[47] Luo, Yuheng, Xu. Fusion of VMD-AR with Adaptive Gaussian Mixture Particle Filtering for Pedestrian Trajectory Tracking[J]. Digital Signal Processing, 2024, 146(1): 104386-

104395.

[48] Pareek, Anshul, Nidhi Arora. Re-projected SURF Features Based Mean-Shift Algorithm For

Visual Tracking[J]. Procedia Computer Science, 2020, 167(1): 1553-1560.

[49] Murzova, Mariya A, Farber. The α-β Filter for Tracking Maneuvering Objects with LFM

Waveforms[C]// 2017 6th International Conference on Engineering and Telecommunication,

IEEE. 2017: 104-107.

[50] Zhang Y, Gill G, Cheng W. Exploring Influential Factors and Endogeneity of Traffic Flow of Different Lanes on Urban Freeways Using Bayesian Multivariate Spatial Models[J]. Journal of Traffic and Transportation Engineering (English Edition), 2023, 10(1): 104-115.

[51] Song, Moon-Huyung, Kim. A Novel Approach to the Enhancement of Lane Estimator by Using Kalman Filter[C]// 2017 17th International Conference on Control, Automation and Systems, IEEE. 2017: 1078-1082.

[52] Chen S, Huang L, Chen H. Multi-Lane Detection and Tracking Using Temporal-Spatial Model and Particle Filtering[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(3): 2227-2245.

[53] Liu S, Lu L, Zhong X. Effective Road Lane Detection and Tracking Method Using Line Segment Detector[C]// 2018 37th Chinese Control Conference, IEEE. 2018: 5222-5227.

[54] Lim K H, Seng K P, Ang L M. Lane Detection and Kalman-Based Linear-Parabolic Lane Tracking[J]. IEEE, 2009: 351-354.

[55] Yin, Fang zhi, Changyin Dong. An Anti-disturbance Lane-changing Trajectory Tracking Control Method Combining Extended Kalman Filter and Robust Tube-based Model Predictive Control[J]. Journal of Intelligent Transportation Systems, 2024, 1(1): 1-16.

[56]Jiao X, Yang D, Jiang K. Real-time Lane Detection and Tracking for Autonomous Vehicle Applications[J]. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, 2019, 233(9): 2301-2311.

[57] Liu G, Worgotter F, Markelic I. Stochastic Lane Shape Estimation Using Local Image Descriptors[J]. IEEE Transactions on Intelligent Transportation Systems, 2013, 14(1):13-21.

[58] Franks U, Loose H, Knoppel C. Lane Recognition on Country Roads[C]// Intelligent Vehicles Symposium, IEEE. 2007: 99-104.

[59] Wang Y, Dahnoun N, Achim A. A Novel System for Robust Lane Detection and Tracking[J]. Signal Processing, 2012, 92(2): 319-334.

[60] Gupta, Any, Choudhary. A Framework for Lane Prediction Based on Vehicle Detection and Tracking[C]// National Conference on Computer Vision, Pattern Recognition, Image Processing, and Graphics, IEEE. 2020: 578-588.

[61]Jang, Baek S H, Park S Y. Curved Lane Detection Using Robust Feature Extraction[C]// The 2014 2nd International Conference on Systems and Informatics, IEEE. 2014: 109-112.

[62] Guo C, Meguro J, Yamaguchi K. Improved Lane Detection Based on Past Vehicle Trajectories[C]// 17th International IEEE Conference on Intelligent Transportation Systems, IEEE. 2014: 1956-1963.

[63] Lin C W, Tseng D C, Wang H Y. A Robust Lane Detection and Verification Method for Intelligent Vehicles[C]// 2009 Third International Symposium on Intelligent Information Technology Application, IEEE. 2009: 521-524.

[64] Hou Y, Ma Z, Liu C. Learning Lightweight Lane Detection CNNs by Self Attention Distillation[C]// Proceedings of the IEEE/CVF International Conference on Computer Vision, IEEE. 2019: 1013-1021.

[65] Paszke A, Chaurasia A, Kim S. Enet: A Deep Neural Network Architecture for Real-time Semantic Segmentation[J]. arXiv preprint arXiv:1606.02147, 2016.

[66] Vaswani A, Shazeer N, Parmar N. Attention is All You Need[J]. Advances in Neural Information Processing Systems, 2017, 30(1), 1-11.

[67] Ko Y, Lee Y, Azam S. Key Points Estimation and Point Instance Segmentation Approach for Lane Detection[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 23(7): 8949-8958.

[68] Li X, Li J, Hu X. Line-cnn: End-to-end Traffic Line Detection with Line Proposal Unit[J]. IEEE Transactions on Intelligent Transportation Systems, 2019, 21(1): 248-258.

[69] Ioffe S, Szegedy C. Batch normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift[C]// International Conference on Machine Learning, IEEE. 2015: 448-456.

[70]Jian S, Kaiming H, Shaoqing R. Deep residual learning for image recognition[C]// IEEE Conference on Computer Vision & Pattern Recognition, IEEE. 2016: 770-778.

[71] Han K, Wang Y, Tian Q. Ghostnet: More Features from Cheap Operations[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE. 2020: 1580-1589.

[72] Wang Q, Wu B, Zhu P. ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE. 2019: 11534-11542.

[73] Yoo S, Lee HS, Myeong H. End-to-end Lane Marker Detection Via Row-wise

Classification[C]// Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops, IEEE. 2020: 1006-1017.

[74] Philion, Jonah. Fastdraw: Addressing the Long Tail of Lane Detection by Adapting a Sequential Prediction Network[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE. 2019: 11582-11591.