基于MEMS激光雷达的车辆目标识别算法

doi:10.12382/bgxb.2021.0822

摘要/Abstract

摘要：

针对传统线阵激光雷达对地面目标识别准确率低的问题,设计一种基于MEMS激光雷达推扫成像点云识别算法。引入直通滤波和栅格分割算法缩减原始点云数据,有效提高算法的运算速度。结合MEMS激光雷达点云有序化处理方法,提出基于数学形态学的点云聚类算法,将去除地面后的点云分割为相互独立的点云簇。在此基础上使用自适应阈值的分布直方图去噪算法,去除点云簇周围的离群噪点。设计多特征复合判据,直接处理聚类去噪后的三维激光点云,实现对目标的准确识别。分析典型实验条件下的数据处理结果,识别准确率达到了94.9%,表明该方法具有良好的泛化能力和准确性。

关键词: MEMS激光雷达, 推扫成像, 车辆, 目标识别

Abstract:

In order to solve the problem of low recognition accuracy of traditional linear array Lidar, recognition algorithm based on MEMS LiDAR pushbroom scanning is designed. To reduce the amount of computation, directly filtering and grid segmentation algorithms are introduced to reduce the amount of original point clouds and effectively improve the real-time performance of detection. Combined with the organized processing method of MEMS LiDAR point cloud, a point cloud clustering algorithm based on mathematical morphology is proposed, which divides the point clouds after removing the ground points into independent point cloud clusters. The denoising algorithm based on distribution histogram with adaptive threshold is used to remove the outlier noise points around the targets. On this basis, a multifeature composite criterion is designed to directly process the three-dimensional LiDAR point clouds after clustering denoising to realize the accurate recognition of the targets. The data processing results under different experimental conditions are analyzed, and the recognition accuracy reaches 94.9%, which shows that the method has good generalization ability and accuracy.

Key words: MEMS LiDAR, pushbroom imaging, vehicle, target recognition

霍健, 陈慧敏, 马云飞, 郭鹏宇, 杨旭, 孟祥盛. 基于MEMS激光雷达的车辆目标识别算法[J]. 兵工学报, 2023, 44(4): 940-948.

HUO Jian, CHEN Huimin, MA Yunfei, GUO Pengyu, YANG Xu, MENG Xiangsheng. Vehicle Target Recognition Algorithm Based on MEMS LiDAR[J]. Acta Armamentarii, 2023, 44(4): 940-948.

图/表 15

图1 弹载MEMS激光雷达工作原理示意图

Fig.1 Schematic diagram of missile borne MEMS LiDAR

图2 等效实验场景图

Fig.2 Equivalent experimental scene

表1 一维MEMS激光雷达性能参数

Table 1 Parameters of one-dimensional MEMS LiDAR

参数	数值
波长/nm	905
最大探测距离/m	75
测距精度/cm	2
视场角/(°)	60
角分辨率/(°)	0.2
帧频/Hz	360

图3 MEMS激光雷达和线阵激光雷达点云对比图

Fig.3 Comparison between MEMS LiDAR point cloud and linear array LiDAR point cloud

图4 直通滤波前后的点云对比图

Fig.4 Comparison of point cloud images before and after directly filtering

图5 点云分割前后对比

Fig.5 Comparison before and after ground points segmentation

图6 无序点云和有序点云的矩阵形式对比

Fig.6 Comparison of matrix formats of unorganized point cloud and organized point cloud

图7 点云有序化处理

Fig.7 Organized processing of point cloud

图8 原始点云图和黑白图对比

Fig.8 Comparison between original point cloud and black-white image

图9 开闭运算后的点云簇位置图

Fig.9 Position diagram of point cloud cluster after morphological operations

图10 聚类运算后提取的独立点云

Fig.10 Independent point cloud extracted after clustering operations

图11 点云目标的分布直方图

Fig.11 Distribution histogram of point cloud targets

表2 点云特征参数

Table 2 Point cloud feature parameters

编号	特征名称	维数
1	点云数量	1
2	点云外框尺寸	3
3	点云归一化坐标中心	3
4	点云空间离散值	6
5	点云三维协方差矩阵	9
6	点云三维协方差矩阵特征值	3

图12 目标识别效果图

Fig.12 Effect drawing of target recognition

表3 实验采集的车辆数量

Table 3 Number of vehicles collected in the experiment

车辆数量及误判数量	种类
车辆数量及误判数量	私家车	公交车	摩托车	厢式车
数量	542	11	43	30
误判数量	26	0	2	4

参考文献 25

[1]	于加其, 杨树兴, 朱伯立. 弹载激光成像雷达距离像的目标提取技术[J]. 北京理工大学学报, 2016, 36(12):1279-1282.
	YU J Q, YANG S X, ZHU B L. Target extraction base on range image from missile-borne imaging LADAR[J]. Transactions of Beijing Institute of Technology, 2016, 36(12): 1279-1282. (in Chinese)
[2]	徐国艳, 牛欢, 郭宸阳, 等. 基于三维激光点云的目标识别与跟踪研究[J]. 汽车工程, 2020, 42(1): 38-46.
	XU G Y, NIU H, GUO C Y, et al. Research on target recognition and tracking based on 3D laser point cloud[J]. Automotive Engineering, 2020, 42(1): 38-46. (in Chinese)
[3]	武军安, 郭锐, 刘荣忠, 等. 末敏弹线阵激光雷达对地面装甲目标的提取方法[J]. 红外与激光工程, 2019, 48(3): 140-147.
	WU J A, GUO R, LIU R Z, et al. Armored target extraction method based on linear array Lidar of terminal sensitive sub-ammunition[J]. Infrared and Laser Engineering, 2019, 48(3): 140-147. (in Chinese)
[4]	党吉圣, 杨军. 多特征融合的三维模型识别与分割[J]. 西安电子科技大学学报, 2020, 47(4): 149-157.
	DANG J S, YANG J. 3D model recognition and segmentation based on muti-feature fusion[J]. Journal of Xidian University, 2020, 47(4): 149-157. (in Chinese)
[5]	冯发杰, 丁亚洲, 吏军平, 等. 使用显著性划分的机载激光雷达点云滤波[J]. 红外与激光工程, 2020, 49(8): 26-34.
	FENG F J, DING Y Z, LI J P, et al. Airborne Lidar point cloud filtering using saliency division[J]. Infrared and Laser Engineering, 2020, 49(8): 26-34. (in Chinese)
[6]	LIU L P, XIAO J, WANG Y B, et al. A novel rock-mass point cloud registration method based on feature line extraction and feature point matching[J]. Transactions on Geoscience and Remote Sensing, 2021, 60: 5701117.
[7]	CHEN G H, JIN Q, ZHANG X Q, et al. Airborne lidar point cloud classification fusing spectral information[C]// Proceedings of the 2021 3rd International Conference on Intelligent Control, Measurement and Signal Processing and Intelligent Oil Field. Xi’an, China: IEEE, 2021:177-184.
[8]	蒋罕寒, 郭锐, 武军安, 等. 高旋掠飞弹载激光雷达稀疏点云插值算法[J]. 弹箭与制导学报, 2021, 41(1): 120-123.
	JIANG H H, GUO R, WU J A, et al. A sparse point cloud interpolation algorithm for the hedgehopping munition with high rotational speed carrying laser radar[J]. Journal of Projectiles, Rockets, Missiles and Guidance, 2021, 41(1): 120-123. (in Chinese)
[9]	刘亚文, 张颖, 陈泉. 结合全景影像的车载街景点云数据增强方法[J]. 武汉大学学报(信息科学版), 2020, 45(7): 1015-1020.
	LIU Y W, ZHANG Y, CHEN Q. Vehicle point cloud data enhancement method combined with panoramic image[J]. Geomatics and Information Science of Wuhan University, 2020, 45(7): 1015-1020. (in Chinese)
[10]	张雨禾, 耿国华, 魏潇然, 等. 采用密度空间聚类的散乱点云特征提取方法[J]. 西安电子科技大学学报, 2017, 44(2): 114-120.
	ZHANG Y H, GENG G H, WEI X R, et al. Feature extraction of point clouds using the DBSCAN clustering[J]. Journal of Xidian University(Natural Science), 2017, 44(2):114-120. (in Chinese)
[11]	LEE S M, IM J J, LEE B H, et al. A real-time grid map generation and object classification for ground-based 3D LIDAR data using image analysis techniques[C]//IEEE International Conference on Image Processing. Hong Kong, China: IEEE, 2010: 2253-2256.
[12]	MILLER C I, THOMAS J J, KIM A M, et al. Application of image classification techniques to multispectral Lidar point cloud data[C]//Proceedings of the SPIE. San Francisco, CA, US:SPIE, 2016: 1001-1012.
[13]	陆军, 华博文, 朱波. 基于局部曲面特征直方图的点云识别[J]. 模式识别与人工智能, 2020, 33(10): 934-943. doi: 10.16451/j.cnki.issn1003-6059.202010008
	LU J, HUA B W, ZHU B. Point cloud recognition based on local surface feature histogram[J]. Pattern Recognition and Artificial Intelligence, 2020, 33(10): 934-943. (in Chinese) doi: 10.16451/j.cnki.issn1003-6059.202010008
[14]	邹子强, 刘星, 张雪峰. 弹载线阵扫描激光雷达的目标识别算法[J]. 探测与控制学报, 2019, 41(1): 43-47.
	ZOU Z Q, LIU X, ZHANG X F. Target recognition algorithm of linear array scanning laser radar[J]. Journal of Detection & Control, 2019, 41(1): 43-47. (in Chinese)
[15]	HUANG R, XU Y S, STILLA U. Extraction of multi-scale geometric features for point cloud classification[C]//Proceedings of 2019 IEEE International Geoscience and Remote Sensing Symposium. Yokohama, Japan: IEEE, 2019:2499-2502.
[16]	ZHAO Z Y, CHENG Y L, SHI X S, et al. Classification of LiDAR point cloud based on multiscale features and PointNet[C]// Proceedings of the 2018 Eighth International Conference on Image Processing Theory, Tools and Applications. Xi’an, China: IEEE, 2018:1-7.
[17]	NI P Y, ZHANG W G, ZHU X X, et al. PointNet++ grasping: learning an end-to-end spatial grasp generation algorithm from sparse point clouds[C]//Proceedings of the 2020 IEEE International Conference on Robotics and Automation. Paris, France: IEEE, 2020: 3619-3625.
[18]	郑壮壮, 曹万科, 邹渊, 等. 动态环境下无人地面车辆点云地图快速重定位方法[J]. 兵工学报, 2020, 41(8):1581-1589. doi: 10.3969/j.issn.1000-1093.2020.08.013
	ZHENG Z Z, CAO W K, ZOU Y, et al. Rapid Localization of unmanned ground vehicles in dynamic environment using point cloud maps[J]. Acta Armamentarii, 2020, 41(8): 1581-1589. (in Chinese) doi: 10.3969/j.issn.1000-1093.2020.08.013
[19]	SHI S S, GUO C X, JIANG L, et al. PV-RCNN: Point-voxel feature set abstraction for 3D object detection[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle, WA, US: IEEE, 2020: 10526-10535.
[20]	CHARLES R Q, SU H, MO K C, et al. PointNet: Deep learning on point sets for 3D classification and segmentation[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI, US: IEEE, 2017: 77-85.
[21]	马振宇, 庞勇, 李增元, 等. 地基激光雷达森林近地面点云精细分类与倒木提取[J]. 遥感学报, 2019, 23(4):743-755.
	MA Z Y, PANG Y, LI Z Y, et al. Fine classification of near-ground point cloud based on terrestrial laser scanning and detection of forest fallen wood[J]. Journal of Remote Sensing, 2019, 23(4): 743-755. (in Chinese)
[22]	KAMENCAY P, SINKO M, HUDEC R, et al. Improved feature point algorithm for 3D point cloud registration[C]//Proceedings of the 2019 42nd International Conference on Telecommunications and Signal Processing. Budapest, Hungary: IEEE, 2019: 517-520.
[23]	WU S, WEN C L, LUO H, et al. Using mobile LiDAR point clouds for traffic sign detection and sign visibility estimation[C]// Proceedings of the 2015 IEEE International Geoscience and Remote Sensing Symposium. Milan, Italy: IEEE, 2015: 565-568.
[24]	ZHOU X H, YANG H X, YANG H. Point cloud grid reduction method based on feature parameters[C]//Proceedings of the 2021 IEEE International Conference on Artificial Intelligence and Industrial Design. Guangzhou, China: IEEE, 2021: 323-326.
[25]	YANG W J, LIU Y, HE H G, et al. Airborne Lidar and photogrammetric point cloud fusion for extraction of urban tree metrics according to street network segmentation[J]. IEEE Access, 2021, 9(7): 97834-97842. doi: 10.1109/ACCESS.2021.3094307 URL