尖轨复杂型面激光测量与三维重构方法研究

doi:10.3901/JME.2024.23.329

摘要/Abstract

摘要： 尖轨是道岔转辙器的关键基础部件，关乎着列车运行平稳性和安全性，其型面测量与重构是实现高效高质自动化加工的根本前提。针对线激光传感器测量所得海量点云数据处理速度慢和配准重构精度低的问题，提出一种尖轨复杂型面的激光测量与三维重构方法。首先，采用组合滤波法去除点云离群噪点，弥补了单一算法去噪能力不足的缺陷，结合直观对比结果和量化评估标准选择双边滤波进行平滑处理，提出基于提取尖轨轮廓的体素网格法完成精简。然后，根据尖轨端面几何特征提出一种尖轨点云三维型面粗配准方法，并对精配准算法改进，实现点云配准，精配准后左右侧点云的均方根误差值分别为0.065 mm和0.067 mm。最后，采用改进的贪婪投影法对配准后的点云进行三维重构，并对比重构数据与实测数据，轨高的平均偏差为0.16 mm，轨底宽的平均偏差为0.17 mm，验证了尖轨复杂型面激光测量与三维重构方法的有效性。

关键词: 激光测量, 三维重构, 配准, 尖轨, 点云

Abstract: Switch rail is the crucial basic part of turnout rutters, which is related to the smoothness and safety of train operation, and the accuracy of its profile measurement and reconstruction is the fundamental prerequisite for the realisation of efficient and high-quality automated machining. Aiming at the problems of slow processing speed and low accuracy of registration reconstruction of massive point cloud data obtained from line laser sensor measurements, it is proposed that a laser measurement and three-dimensional reconstruction method for complex surface of pointed rail. Firstly, the combined filtering method is used to remove the point cloud outlier noise, which makes up for the deficiency of the single algorithm's denoising ability, combining the intuitive comparison results and quantitative evaluation criteria to select bilateral filtering for smoothing, and proposed the voxel mesh method based on the extraction of the switch rail profile to complete streamlining. Then, according to the geometrical characteristics of the switch rail, a switch rail point cloud coarse registration algorithm is proposed, and the iterative closest point fine registration algorithm is improved to achieve registration, and the RMSE values of the left and right side of the point cloud after fine registration are 0.065 mm and 0.067 mm, respectively. Finally, the improved greedy projection method is used to reconstruct the registration point cloud in three dimensions, and comparing the reconstructed data with the measured data, the average deviation of the rail height is 0.16 mm, and the average deviation of the rail bottom width is 0.17 mm, which verifies the validity of the laser measurement and three-dimensional reconstruction method of the complex profile of the switch rail.

Key words: laser measurement, three-dimensional reconstruction, registration, switch rail, point cloud

中图分类号:

TG156

樊文刚, 吴昌昕, 吴志伟, 李江. 尖轨复杂型面激光测量与三维重构方法研究[J]. 机械工程学报, 2024, 60(23): 329-340.

FAN Wengang, WU Changxin, WU Zhiwei, LI Jiang. Research on Laser Measurement and Three-dimensional Reconstruction Methods for Complex Profiles of Switch Rail[J]. Journal of Mechanical Engineering, 2024, 60(23): 329-340.

参考文献

[1] 司道林，赵振华，王树国，等. 新型50 kg/m钢轨9号道岔尖轨转换特性研究[J]. 中国铁道科学，2021，42(4)：27-33. SI Daolin，ZHAO Zhenhua，WANG Shuguo，et al. Research on the conversion characteristics of tip rail for new 50 kg/m steel rail 9# turnout[J]. China Railway Science, 2021, 42(4):27-33.
[2] 王璞，曾瑞东，王树国. 高速铁路道岔尖轨不足位移控制方法[J]. 交通运输工程学报，2022，22(2)：87-98. WANG Pu，CENG Ruidong，WANG Shuguo. Control method for insufficient displacement of switch rail in high-speed railway turnout[J]. Journal of Traffic and Transportation Engineering，2022，22(2)：87-98.
[3] 司道林，赵振华，杨东升，等. 高速道岔尖轨断裂对道岔状态影响的试验研究[J]. 铁道工程学报， 2020，37(9)：18-22，28. SI Daolin，ZHAO Zhenhua，YANG Dongsheng，et al. Experimental research on the influence of switch rail fracture on high-speed turnout status[J]. Journal of Railway Engineering Society，2020，37(9)：18-22.
[4] 樊文刚，刘月明，李建勇. 高速铁路钢轨打磨技术的发展现状与展望[J]. 机械工程学报，2018，54(22)：184-193. FAN Wengang，LIU Yueming，LI Jianyong. Development status and prospect of rail grinding technology for high-speed railway [J]. Journal of Mechanical Engineering. 2018，54(22)：184-193.
[5] FAN W G，WU C X，WU Z W，et al. Static contact mechanism between serrated contact wheel and rail in rail grinding with abrasive belt[J]. Journal of Manufacturing Processes. 2022，84：1229-1245.
[6] 丁建军，卢光超，孙林，等基于扫描点云的叶片特征型线分析方法及实现[J]. 机械工程学报，2022，58(17)：41-48. DING Jianjun，LU Guangchao，SUN Lin，et al. Method and implementation of blade feature profile analysis based on scanning point cloud[J]. Journal of Mechanical Engineering，2022，58(17)：41-48.
[7] 李涛涛，杨峰，许献磊. 基于多视觉线结构光传感器的大尺度测量方法[J]. 中国激光，2017，44(11)：130-140. LI Taotao，YANG Feng，XU Xianlei. Large-scale measurement method based on multi-vision line structured optical sensor[J]. China Laser，2017，44(11)：130-140.
[8] 韩鸿烈. 基于双目视觉的尖轨轮廓检测系统研究[D]. 兰州：兰州交通大学，2022. HAN Honglie. Research on tip rail profile detection system based on binocular vision [D]. Lanzhou：Lanzhou Jiaotong University，2022.
[9] 高文洲. 基于机器视觉的高速道岔尖轨检测系统设计[D]. 兰州：兰州交通大学，2021. GAO Wenzhou. Design of high-speed turnout tip track detection system based on machine vision [D]. Lanzhou：Lanzhou Jiaotong University，2021
[10] 罗磊. 图像处理技术在尖轨检测中的应用研究[D]. 成都：西南交通大学，2018. LUO Lei. Research on the application of image processing technology in the detection of sharp rail [D]. Chengdu:Southwest Jiaotong University，2018.
[11] 孙爱程. 基于双目视觉的尖轨轮廓检测系统的设计与实现[D]. 武汉：武汉理工大学，2016. SUN Aicheng. Design and realisation of tip track profile detection system based on binocular vision [D]. Wuhan：Wuhan University of Technology，2016.
[12] 武福，豆辉，武云，等. 基于机器视觉的尖轨轮廓检测方法[J]. 光学技术，2020，46(4)：453-460. WU Fu，DU Hui，WU Yun，et al. A machine vision-based method for tip-track profile detection[J]. Optical Technology，2020，46(4)：453-460.
[13] 郑小江. 60钢轨和9号道岔尖轨光学测量精度及廓形参数研究[D]. 成都：西南交通大学，2019. ZHENG Xiaojiang. Research on optical measurement accuracy and contour parameters of 60 rails and No.9 turnout tip rails [D]. Chengdu：Southwest Jiaotong University，2019.
[14] 袁富国. 基于面结构光的高速道岔尖轨检测系统研究[D]. 兰州：兰州交通大学，2022. YUAN Fuguo. Research on high-speed turnout tip track detection system based on surface structured light [D]. Lanzhou：Lanzhou Jiaotong University，2022.
[15] 刘永刚，于丰宁，章新杰，等. 基于激光点云与图像融合的3D目标检测研究[J]. 机械工程学报， 2022，58(24)：289-299. LIU Yonggang，YU Fengning，ZHANG Xinjie，et al. Research on 3D target detection based on laser point cloud and image fusion[J]. Journal of Mechanical Engineering，2022，58(24)：289-299.
[16] 王世丰. 基于线激光传感器的心轨型面测量方法研究[D]. 北京：北京交通大学，2022. WANG Shifeng. Research on the measurement method of heart track profile based on line laser sensor [D]. Beijing：Beijing Jiaotong University，2022.
[17] ZHANG F，ZHANG C，YANG H，et al. Point cloud denoising with principal component analysis and a novel bilateral filter[J]. Traitement du Signal，2019，36(5)：393-398.
[18] 陈强，黄声享，王韦. 小波去噪效果评价的另一指标[J]. 测绘信息与工程，2008，33(5)：13-14. CHEN Qiang，HUANG Shengxiang，WANG Wei. Another index for the evaluation of wavelet denoising effect[J]. Surveying and Mapping Information and Engineering，2008，33(5)：13-14.
[19] 张彬，熊传兵. 基于体素下采样和关键点提取的点云自动配准[J]. 激光与光电子学进展，2020，57(4)：101-109. ZHANG Bin，XIONG Chuanbing. Automatic point cloud alignment based on voxel downsampling and keypoint extraction[J]. Advances in Lasers and Optoelectronics，2020，57(4)：101-109.
[20] 宋涛，曹利波，赵明富，等. 三维点云中关键点的配准与优化算法[J]. 激光与光电子学进展， 2021，58(4)：367-375. SONG Tao，CAO Libo，ZHAO Mingfu，et al. Alignment and optimisation algorithm for key points in 3D point clouds[J]. Advances in Lasers and Optoelectronics，2021，58(4)：367-375.
[21] AIGER D，MITRA N J，COHEN-OR D. 4-points congruent sets for robust pairwise surface registration[J]. ACM Transactions on Graphics， 2008，27(3)：1-10.
[22] MELLADO N，AIGER D，MITRA N J. Super 4PCS fast global pointcloud registration via smart indexing[J]. Computer Graphics Forum. 2014，33(5)：205-215.
[23] RUSU R B，BLODOW N，BEETZ M. Fast point feature histograms (FPFH) for 3D registration[C]//IEEE，2009.
[24] 邱春丽. 基于点云的曲面重建技术研究[D]. 北京：北京交通大学，2014. QIU Chunli. Research on surface reconstruction technology based on point cloud [D]. Beijing ：Beijing Jiaotong University，2014.
[25] QIAN N. Efficient poisson-based surface reconstruction of 3D model from a non-homogenous sparse point cloud[M]. Cham：Springer International Publishing，578-585.
[26] LIU J，BAI D，CHEN L. 3-D Point cloud registration algorithm based on greedy projection triangulation[J]. Applied Sciences，2018，8(10)：1776.