基于机器视觉的吊弦动态抬升量测量方法  

Measurement Method for Dynamic Lifting of Dropper Based on Machine Vision

作　　者：阮杰[1] 龙鹏 许一统 李红梅[2] RUAN Jie;LONG Peng;XU Yitong;LI Hongmei(Hubei Key Laboratory of Advanced Technology for Automotive Components,Wuhan University of Technology,Wuhan Hubei 430070,China;Electrification Design and Research Institute,China Railway Siyuan Survey and Design Group Co.,Ltd.,Wuhan Hubei 430063,China)

机构地区：[1]武汉理工大学现代汽车零部件技术湖北省重点实验室,湖北武汉430070 [2]中铁第四勘察设计院集团有限公司电气化设计研究院,湖北武汉430063

出　　处：《中国铁道科学》2022年第2期143-151,共9页China Railway Science

基　　金：中国国家铁路集团有限公司科技研究开发计划课题(N2019G048)。

摘　　要：为研究吊弦的疲劳寿命问题须先对吊弦的动态抬升量进行定点实时测量,因此提出一种基于机器视觉的测量方法。首先,通过鼠标交互的方式,在输入视频流的第1帧图像上框选吊弦在接触线侧线夹处的图像作为初始模板图像,并将其中心点定义为初始测量点,以此确定测量的定位。在此基础上,基于二维离散余弦变换将模板图像的信息从图像域转换到频率域,选取其中低频信号代表的系数矩阵,通过哈希算法将该矩阵转换为哈希值,并对输入的每帧图像均采用模板匹配方法,实现对测量点的跟踪,提取测量点的像素坐标。然后,根据相机成像原理,建立吊弦抬升量测量模型,通过摄像机内参标定和坐标系逆变换推导出像素坐标与实际位移之间的转换公式,完成对吊弦动态抬升量的计算。为验证测量方法的有效性,选取京广高速铁路武汉至乌龙泉东区间段进行现场测试。结果表明:采用该方法处理单帧图像的时间仅需0.1 s,吊弦动态抬升量测量误差最大为4.6%,具有较高的测量精度,能够实时、有效地获得吊弦动态抬升量。In order to study the fatigue life of the dropper,the dropper dynamic lifting of the fixed point in real time should be measured primarily.Therefore,a measuring method based on machine vision is proposed.Firstly,by means of mouse interaction,the image of the dropper at the side clip of the contact line is selected as the initial template image on the first frame image of the input video stream,and its center point is defined as the initial measurement point to determine the location of the measurement.On this basis,the information of the template image is converted from the image domain to the frequency domain based on the twodimensional discrete cosine transform.The coefficient matrix represented by the lowfrequency signal is selected,and the matrix is converted into hash value by the hash algorithm.The template matching method is adopted for each frame of the input image to realize the tracking of the measurement points and extract the pixel coordinates of the measurement points.Then,according to the imaging principle of the camera,a measurement model of the dropper lifting is established,and the conversion formula between the pixel coordinates and the actual displacements is deduced through the calibration of the camera internal parameters and the inverse transformation of the coordinate system.The calculation of the dynamic lifting of the dropper is completed.In order to verify the validity of the measurement method,the section from Wuhan to Wulongquan east of BeijingGuangzhou high speed railway is selected for field testing.The results show that the method takes only 0.1 s to process a single frame of image,and the maximum measurement error of the dropper dynamic lifting is 4.6%.It has high measurement accuracy,and can effectively obtain the dropper dynamic lifting in real time.

关 键 词：机器视觉 吊弦动态抬升量 哈希算法 图像处理 

分 类 号：U225.48[交通运输工程—道路与铁道工程]