基于机器视觉的火车轮对轴端标记自动识别算法研究  

Research on automatic recognition algorithm of axle end mark of train wheelset based on machine vision

作　　者：刘德志 曾勇[1] 袁雨鑫 卢倩[1] 刘聪[1] LIU Dezhi;ZENG Yong;YUAN Yuxin;LU Qian;LIU Cong(Yancheng Institute of Technology,Yancheng 224007,China)

机构地区：[1]盐城工学院,盐城224007

出　　处：《现代制造工程》2022年第7期113-120,共8页Modern Manufacturing Engineering

基　　金：国家自然科学基金资助项目(51405418);江苏省高校自然科学基金重大资助项目(18KJA460009);江苏省“六大高峰”人才资助项目(JXQC-028);江苏省“青蓝工程”人才资助项目(2021)。

摘　　要：为了对火车轮对轴端标记进行识别录入,基于机器视觉技术提出了一种针对轮对轴端的熔炼号、钢种号、单位号、顺序号、年月、轴型标记及方位标记等7种类型标记的整体自动识别算法。在对获取的轴端图像进行灰度处理、图像增强及滤波处理等操作的基础上,采用圆检测和双线性插值相结合的方法,解决了轴端标记各角度的倾斜校正问题。通过阈值分割和形态学处理提取目标区域,并采用投影法对目标进行行分割,最后采用模板匹配法对目标字符进行识别。通过构建识别系统实验表明,该算法能有效针对轴端标记进行识别分类,准确率为97%,识别速度为每幅5 s。In order to recognize and input the axle end mark of train wheelset,based on the machine vision technology,an integral automatic recognition algorithm was proposed for seven types of marks,including smelting number,steel grade number,unit number,sequence number,year and month,shaft type mark and azimuth mark,at the axle end of wheelset.Based on the gray processing,image enhancement and filter processing of the acquired image of the axle end,the problem of tilt correction of each angle of the mark of the axle end was solved by the method of combining circle detection and bilinear interpolation.The target region was extracted by threshold segmentation and morphology processing,and the line segmentation was performed by projection method.Finally,the target character was recognized by template matching method.The experimental results show that the algorithm can effectively identify and classify the axle end marks with an accuracy of 97%and a recognition speed of 5 s per image.

关 键 词：机器视觉 火车轮对 标记识别 倾斜校正 

分 类 号：TH164[机械工程—机械制造及自动化]