作者：伍川辉1,2, 邓越1,2, 于涛1,2, 陈国俊1,2, 李恒奎3
作者单位：1. 西南交通大学机械工程学院，四川 成都 610031;
2. 轨道交通运维技术与装备四川省重点实验室，四川 成都 610031;
3. 青岛四方机车车辆股份有限公司，山东 青岛 266031
关键词：受电弓滑板;全局标定;线结构光;多视觉传感器
摘要：
受电弓磨耗缺陷作为受电弓最主要故障缺陷，但由于碳滑板横向尺度大、边缘精度低等特点，单个视觉传感器不能覆盖到所有受电弓轮廓，因此提出一种利用多视觉传感器进行全局标定的标定方法。首先，使用二维靶标对单个视觉传感器进行标定得到摄像机内部参数，然后对激光平面进行标定和求解，最后把三维靶标置于相邻两个摄像机的交叉视角范围内，并求解出两个相邻相机的外部参数。综合摄像机内外部参数和光平面方程即可建立统一坐标系下的数学测量模型，当得到线结构光测量模型后将二维图像的特征点转换为空间点云信息即可实现对滑板磨耗的测量。该系统通过测量三维靶标的尺寸并和真实距离进行对比来验证测量模型的准确性。对比结果表明该方法标定精度高、实现简单、结构稳定，能够有效用来对受电弓碳滑板磨耗的测量。
Research on the calibration method of multi-vision sensors in the wear detection of pantograph skateboard
WU Chuanhui1,2, DENG Yue1,2, YU Tao1,2, CHEN Guojun1,2, LI Hengkui3
1. School of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610031, China;
2. Technology and Equipment of Rail Transit Operation and Maintenance Key Laboratory of Sichuan Province, Chengdu 610031, China;
3. Qingdao Sifang Rolling Stock Co., Ltd., Qingdao 266031 , China
Abstract: The pantograph wear defect is the main fault defect of the pantograph. However, due to the large lateral scale and low edge accuracy of the carbon slide plate, a single vision sensor cannot cover all the pantograph contours. Therefore, a method of using multiple vision sensors is proposed. The calibration method of global calibration. First, use a two-dimensional target to calibrate a single vision sensor to obtain the internal parameters of the camera, and then calibrate and solve the laser plane. Finally, place the three-dimensional target within the cross-view range of two adjacent cameras, and solve the two phases. The external parameters of the adjacent camera. Combining the internal and external parameters of the camera and the light plane equation can establish a mathematical measurement model under a unified coordinate system. When the linear structured light measurement model is obtained, the feature points of the two-dimensional image can be converted into spatial point cloud information to realize the measurement of the skateboard wear. Finally, the system verifies the accuracy of the measurement model by measuring the size of the three-dimensional target and comparing it with the real distance. The comparison results show that this method has high calibration accuracy, simple implementation, stable structure, and can be effectively used to measure the wear of the pantograph carbon sliding plate.
Keywords: pantograph skateboard;global calibration;line structured light;multiple vision sensors
2023, 49(1):7-13 收稿日期: 2021-10-20;收到修改稿日期: 2021-12-27
基金项目: 国家重点研发计划（2020YFB1200300ZL）；成都市重点研发支撑计划（2019-YF05-02685-SN）
作者简介: 伍川辉(1964-),男,四川成都市人,副教授,硕士,研究方向为先进测控技术及应用
参考文献
[1] 张新龙. 基于单目视觉激光扫描成像的受电弓磨损检测系统设计[D]. 石家庄: 石家庄铁道大学, 2019.
[2] 张辉. 双目视觉受电弓表面三维重建技术研究[D]. 成都: 西南交通大学, 2014.
[3] CHOI K H, KIM Y, KIM C. Analysis of fish-eye lens camera self-calibration.[J]. Sensors (Basel, Switzerland), 2019, 19(5): 285-294
[4] ZOU W, WEI Z Z, LIU F L. High-accuracy calibration of line-structured light vision sensors using a plane mirror.[J]. Optics Express, 2019, 27(24): 498-510
[5] LI Y C, FANG S P. Large object vision measurement system based on a fixed connection between a camera and a total station telescope.[J]. Applied Optics, 2019, 58(26): 707-713
[6] 胡浩, 梁晋, 唐正宗, 等. 大视场多像机视频测量系统的全局标定[J]. 光学精密工程, 2012, 20(2): 369-378
[7] 高金刚, 刘智勇, 王华, 等. 基于激光跟踪仪的多目视觉全局标定算法研究[J]. 制造技术与机床, 2020(6): 157-162
[8] 鲁亚楠, 万子敬, 王向军. 一种无公共视场相机位置关系的求解方法[J]. 应用光学, 2017, 38(3): 400-405
[9] 潘华伟, 杨振先, 高春鸣, 等. 一种基于平面模板的多摄像机标定方法[J]. 计算机应用研究, 2011, 28(11): 4357-4360
[10] 张曦, 张健. 线结构光标定方法综述[J]. 激光与光电子学进展, 2018, 55(2): 7-17
[11] 于龙龙, 李艳文, 栾英宝, 等. 基于二维平面靶标的线结构光标定[J]. 仪器仪表学报, 2020, 41(6): 124-131
[12] 李佳莹, 罗哉, 江文松, 等. 基于交比不变性的分区域相机畸变矫正[J]. 中国测试, 2020, 46(9): 125-130
[13] 邾继贵, 于之靖. 视觉测量原理与方法[M]. 北京: 机械工业出版社, 2011.
