基于激光雷达的三维目标检测是自动驾驶领域中的一项关键技术，能准确获取汽车、行人、骑行者等多目标的空间位置和朝向信息，为实现自动驾驶提供可靠性决策，提升三维目标检测精度对进一步提升行车安全具有重要意义。论文主要研究内容如下：

（1）基于ECA Modules-PointPillars的三维目标检测算法。针对PointPillars算法中对点云进行立柱划分存在信息丢失问题，将ECA模块串联在原始下采样模块Conv后重新构建骨干网络，实现对伪图像中位置特征信息的增强和背景噪声等不相关特征信息的弱化。实验结果表明，改进算法与PointPillars、F-PointNet、VoxelNet和SECOND相比，三维模式下汽车的AP分别提高了1.97%、6.57%、11.85%和3.3%。

（2）基于RGB-D-ECA Modules-PointPillars的三维目标检测算法。基于ECA Modules-PointPillars的三维目标检测算法在汽车和骑行者的检测上已取得显著成效，为了进一步优化行人类别的检测性能并增强模型对复杂场景的适应能力，融合点云与二维图像同时在Pillar Feature Net中引入平均池化并对ECA模块做出适应性改进。实验结果表明，改进算法与ECA Modules-PointPillars算法相比，鸟瞰图模式、三维模式和AOS模式下行人的AP分别提高了3.65%、4.17%和6.15%。

（3）基于SE-FC-Voxel RCNN的三维目标检测算法。为了进一步提高点云处理的维度和深度，以适应更复杂多变的交通环境，从2D到3D卷积，结合Voxel RCNN算法，在3D骨干网络中引入改进过的焦点稀疏卷积Focals ConvNet-F模块。实验结果表明，改进算法与Voxel RCNN和FC-Voxel RCNN相比，鸟瞰图模式和三维模式下AP分别提高0.84%、6.36%和0.08%、5.63%。

Three-dimensional object detection based on LiDAR is a key technology in the field of automatic driving, which can accurately obtain the spatial position and orientation information of multiple targets such as cars, pedestrians, cyclists, etc., and provide reliable decision-making for the realization of automatic driving, and the improvement of the accuracy of three-dimensional object detection is of great significance to further enhance driving safety. The main research contents of the thesis are as follows:

（1）3D object detection algorithm based on ECA Modules-PointPillars. Aiming at the problem of information loss in the PointPillars algorithm for the column division of the point cloud, the backbone network is reconstructed by connecting the ECA modules in series after the original downsampling module Conv, so as to realize the enhancement of the positional feature information in the pseudo-image and the weakening of the irrelevant feature information such as the background noise. The experimental results show that the improved algorithm improves the AP of the car in 3D mode by 1.97%, 6.57%, 11.85% and 3.3% compared to PointPillars, F-PointNet, VoxelNet and SECOND, respectively.

（2）3D object detection algorithm based on RGB-D-ECA Modules-PointPillars. The 3D object detection algorithm based on ECA Modules-PointPillars has achieved significant results in the detection of cars and cyclists, in order to further optimize the detection performance of the pedestrian category and enhance the model's ability to adapt to complex scenes, the fusion of the point cloud and the 2D image at the same time in the Pillar Feature Net to introduce the average pooling and make adaptive improvements to the ECA module. The experimental results show that the improved algorithm improves the AP of pedestrians by 3.65%, 4.17%, and 6.15% in BEV mode, 3D mode, and AOS mode, respectively, compared to the ECA Modules-PointPillars algorithm.

（3）3D object detection algorithm based on SE-FC-Voxel RCNN. In order to further improve the dimension and depth of point cloud processing for more complex and variable traffic environments, from 2D to 3D convolution, a modified focal sparse convolution module Focals ConvNet-F is introduced into the 3D backbone network by combining the Voxel RCNN algorithm. The experimental results show that the improved algorithm improves AP by 0.84%, 6.36% and 0.08%, 5.63% in BEV mode and 3D mode, respectively, compared with Voxel RCNN and FC-Voxel RCNN.

[1]Meyer M, Kuschk G. Automotive radar dataset for deep learning based 3d object detection[C]//2019 16th european radar conference (EuRAD). IEEE, 2019: 129-132.

[2]黄哲, 王永才, 李德英. 3D目标检测方法研究综述[J]. 智能科学与技术学报, 2023, 5(01): 7-31.

[3]范晶晶, 王力, 褚文博, 等. 基于KDTree树和欧式聚类的越野环境下行人识别的研究[J]. 汽车工程, 2019, 41(12): 1410-1415.

[4]蓝志鹏, 陈锐, 蓝贤桂. 基于K-Means的铁路货运车辆异物识别方法[J]. 机电工程技术, 2023, 52(04): 25-29.

[5]Behley J, Steinhage V, Cremers A B. Laser-based segment classification using a mixture of bag-of-words[C]//2013 IEEE/RSJ International Conference on Intelligent Robots and Systems. IEEE, 2013: 4195-4200.

[6]张长勇, 陈治华, 韩梁. 基于改进DBSCAN的激光雷达障碍物检测[J]. 激光与光电子学进展, 2021, 58(24): 451-458.

[7]张长勇, 韩梁. 基于优化DBSCAN的激光雷达障碍物检测[J]. 激光与光电子学进展, 2022, 59(12): 516-524.

[8]Deng X, Tang G, Wang Q. A novel fast classification filtering algorithm for LiDAR point clouds based on small grid density clustering[J]. Geodesy and Geodynamics, 2022, 13(1): 38-49.

[9]汪世财, 谈东奎, 谢有浩, 等. 基于激光雷达点云密度特征的智能车障碍物检测与跟踪[J]. 合肥工业大学学报(自然科学版), 2019, 42(10): 1311-1317.

[10]王太学, 江智, 江德港, 等. 融合PCA的改进ICP激光点云配准算法[J]. 遥感信息, 2022, 37(02): 70-76.

[11]陈义, 王勇, 李金龙, 等. 基于主成分分析的高效点云配准算法[J]. 激光与光电子学进展, 2023, 60(14): 376-383.

[12]Guo J, Wang G, Guan W, et al. A feasible region detection method for vehicles in unstructured environments based on PSMNet and improved RANSAC[J]. Multimedia Tools and Applications, 2023, 82(28): 43967-43989.

[13]李佳奇, 聂婷, 毕国玲, 等. 基于改进RANSAC算法的雾天自动驾驶汽车视觉图像配准方法[J]. 激光杂志, 2023, 44(11): 54-59.

[14]陈传毅, 罗印升. 基于SHOT、PPF特征的遮挡目标识别研究[J]. 激光杂志, 2021, 42(05): 129-132.

[15]赵毅强, 艾西丁·艾克白尔, 陈瑞, 等. 基于体素化图卷积网络的三维点云目标检测方法[J]. 红外与激光工程, 2021, 50(10): 281-289.

[16]Li B, Zhang T, Xia T. Vehicle Detection from 3D Lidar Using Fully Convolutional Network[J]. arXiv e-prints, 2016: arXiv: 1608.07916.

[17]Yang B, Luo W, Urtasun R. Pixor: Real-time 3d object detection from point clouds[C]//Proceedings of the IEEE conference on Computer Vision and Pattern Recognition. 2018: 7652-7660.

[18]Beltrán J, Guindel C, Moreno F M, et al. Birdnet: a 3d object detection framework from lidar information[C]//2018 21st International Conference on Intelligent Transportation Systems(ITSC). IEEE, 2018: 3517-3523.

[19]Simon M , Milz S , Amende K ,et al. Complex-YOLO: An Euler-Region-Proposal for Real-Time 3D Object Detection on Point Clouds[C]//Computer Vision-ECCV 2018 Workshops: Munich, Germany, September 8-14, 2018, Proceedings, Part I.:Springer, 2019:197-209.

[20]Redmon J, Farhadi A. YOLO9000: better, faster, stronger[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 7263-7271.

[21]Qi C R, Su H, Mo k, et al. PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation[C]//Proceedings of the 30th IEEE Conference on Computer Vision and Pattern Recognition. 2017: 652-660.

[22]QI C R, YI L, SU H, et al. PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space[C]//Annual Conference on Neural Information Processing Systems(NIPS), 2017: 5100-5109.

[23]Qi C R, Liu W, Wu C, et al. Frustum pointnets for 3d object detection from rgb-d data[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 918-927.

[24]王涛, 王文举, 蔡宇. 基于深度学习的三维点云语义分割方法研究[J]. 计算机工程与应用, 2021, 57(23): 18-26.

[25]Shi S, Wang X, Li H. Pointrcnn: 3d object proposal generation and detection from point cloud[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019: 770-779.

[26]Yang Z, Sun Y, Liu S, et al. 3dssd: Point-based 3d single stage object detector[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 11040-11048.

[27]Zhou Y, Tuzel O. Voxelnet: End-to-end learning for point cloud based 3d object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 4490-4499.

[28]Deng J, Shi S, Li P, et al. Voxel r-cnn: Towards high performance voxel-based 3d object detection[C]//Proceedings of the AAAI Conference on Artificial Intelligence. 2021, 35(2): 1201-1209.

[29]Yan Yan, Mao Yuxing. SECOND: Sparsely Embedded Convolutional Detection[J]. Sensors (Basel, Switzerland), 2018, 18(10): 3337.

[30]LANG A H, VORA S, CAESAR H, et al. PointPillars: Fast Encoders for Object Detection from Point Clouds[C]//Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019: 12697-12705.

[31]Engelcke M, Rao D, Wang D Z, et al. Vote3deep: Fast object detection in 3d point clouds using efficient convolutional neural networks[C]//2017 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2017: 1355-1361.

[32]陈德江, 余文俊, 高永彬. 基于改进PointPillars的激光雷达三维目标检测[J]. 激光与光电子学进展, 2023, 60(10): 447-453.

[33]詹为钦, 倪蓉蓉, 杨彪. 基于注意力机制的PointPillars+三维目标检测[J]. 江苏大学学报(自然科学版), 2020, 41(03): 268-273.

[34]田枫, 刘超, 刘芳, 等. 基于改进PointPillars的激光雷达三维目标检测[J]. 激光与光电子学进展, 2024, 61(08): 235-244.

[35]Liu Z, Mao H, Wu C Y, et al. A convnet for the 2020s[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2022: 11976-11986.

[36]Ye M, Xu S, Cao T. Hvnet: Hybrid voxel network for lidar based 3d object detection[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 1631-1640.

[37]Shi S, Guo C, Jiang L, et al. Pv-rcnn: Point-voxel feature set abstraction for 3d object detection[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 10529-10538.

[38]Li J, Sun Y, Luo S, et al. P2v-rcnn: Point to voxel feature learning for 3d object detection from point clouds[J]. IEEE Access, 2021, 9: 98249-98260.

[39]He C, Zeng H, Huang J, et al. Structure aware single-stage 3d object detection from point cloud[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 11873-11882.

[40]Song N, Jiang T, Yao J. JPV-Net: Joint Point-Voxel Representations for Accurate 3D Object Detection[C]//Proceedings of the AAAI Conference on Artificial Intelligence. 2022, 36(2): 2271-2279.

[41]张新钰, 邹镇洪, 李志伟, 等. 面向自动驾驶目标检测的深度多模态融合技术[J]. 智能系统学报, 2020, 15(04): 758-771.

[42]CHEN X, MA H, WAN J, et al. Multi-View 3D object detection network for autonomous driving[C]//Proceeding of the 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, HI, USA: IEEE, 2017:6526-6534.

[43]Ku J, Mozifian M, Lee J, et al. Joint 3d proposal generation and object detection from view aggregation[C]//2018 IEEE/RSJ International Conference on Intelligent Robots and Systems(IROS). IEEE, 2018: 1-8.

[44]Vora S, Lang A H, Helou B, et al. Pointpainting: Sequential fusion for 3d object detection[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 4604-4612.

[45]Pang S, Morris D, Radha H. CLOCs: Camera-LiDAR object candidates fusion for 3D object detection[C]//2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2020: 10386-10393.

[46]梁振明, 黄影平, 宋卓恒, 等. 自动驾驶中基于深度学习的3D目标检测方法综述[J]. 上海理工大学学报, 2024, 46(02): 103-119.

[47]Ioffe S, Szegedy C. Batch normalization: Accelerating deep network training by reducing internal covariate shift[C]//International conference on machine learning. pmlr, 2015: 448-456.

[48]Nair V, Hinton G E. Rectified linear units improve restricted boltzmann machines[C]//Proceedings of the 27th international conference on machine learning(ICML-10). 2010: 807-814.

[49]Liu W, Anguelov D, Erhan D, et al. Ssd: Single shot multibox detector[C]//Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part I 14. Springer International Publishing, 2016: 21-37.

[50]Everingham M, Van Gool L, Williams C K I, et al. The pascal visual object classes(voc) challenge[J]. International journal of computer vision, 2010, 88: 303-338.

[51]Lin T Y, Goyal P, Girshick R, et al. Focal loss for dense object detection[C]//Proceedings of the IEEE international conference on computer vision. 2017: 2980-2988.

[52]Graham B, Engelcke M, Van Der Maaten L. 3d semantic segmentation with submanifold sparse convolutional networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 9224-9232.

[53]Shi S, Wang Z, Shi J, et al. From points to parts: 3d object detection from point cloud with part-aware and part-aggregation network[J]. IEEE transactions on pattern analysis and machine intelligence, 2020, 43(8): 2647-2664.

[54]Li B, Ouyang W, Sheng L, et al. Gs3d: An efficient 3d object detection framework for autonomous driving[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019: 1019-1028.

[55]Geiger A, Lenz P, Stiller C, et al. Vision meets robotics: The kitti dataset[J]. The International Journal of Robotics Research, 2013, 32(11): 1231-1237.

[56]Geiger A, Lenz P, Urtasun R. Are we ready for autonomous driving? the kitti vision benchmark suite[C]//2012 IEEE conference on computer vision and pattern recognition. IEEE, 2012: 3354-3361.

[57]周燕, 许业文, 蒲磊, 等. 自动驾驶场景下的图像三维目标检测研究进展[J/OL]. 计算机科学, 2024-05-25.

[58]王国军. 结构化道路下基于激光雷达的三维检测关键技术研究[D]. 吉林大学, 2021.

[59]Wang Q, Wu B, Zhu P, et al. ECA-Net: Efficient channel attention for deep convolutional neural networks[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 11534-11542.

[60]王忠全. 基于深度学习的目标检测及路径规划研究[D]. 哈尔滨工业大学, 2021.

[61]车爱博. 复杂交通环境下基于点云场景的三维目标检测研究[D]. 长沙理工大学, 2022.

[62]童康, 吴一全. 基于深度学习的小目标检测基准研究进展[J]. 电子学报, 2024, 52(03): 1016-1040.

[63]周燕, 许业文, 蒲磊, 等. 自动驾驶场景下的图像三维目标检测研究进展[J/OL]. 计算机科学, 2024-05-25.

[64]付苗苗, 邓淼磊, 张德贤. 基于深度学习和Transformer的目标检测算法[J]. 计算机工程与应用, 2023, 59(01): 37-48.

[65]Chen Y, Li Y, Zhang X, et al. Focal sparse convolutional networks for 3d object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022: 5428-5437.

[66]赵佳琦, 周勇, 何欣, 等. 基于深度学习的点云分割研究进展分析[J]. 电子与信息学报, 2022, 44(12): 4426-4440.

[67]Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 7132-7141.

[68]张燕咏, 张莎, 张昱, 等. 基于多模态融合的自动驾驶感知及计算[J]. 计算机研究与发展, 2020, 57(09): 1781-1799.