基于机器视觉的螺纹牙型角检测方法研究  

Research on Detecting Method of Thread Included Angle Based on Machine Vision

作　　者：韩宗旺 张伟[1] 程祥[1] 李锦超 高军[1] 李阳 Han Zongwang;Zhang Wei;Cheng Xiang;Li Jinchao;Gao Jun;Li Yang(School of Mechanical Engineering,Shandong University of Technology,Zibo,Shandong 255049,China)

机构地区：[1]山东理工大学机械工程学院,山东省淄博市255049

出　　处：《工具技术》2022年第5期102-106,共5页Tool Engineering

基　　金：教育部产学合作协同育人项目(202101187006);山东省自然科学基金(ZR2020ME157,ZR2016FL15,E5221823A)。

摘　　要：针对螺纹牙型角测量过程中自动化程度和检测效率低等问题,设计了一种基于机器视觉的螺纹牙型角测量系统,开发了螺纹检测平台控制算法和图像处理算法。该系统采用工业相机获取螺纹图像,利用边缘增强的Otsu算法分割螺纹工件图像,基于分割图像的LSD算法检测牙型角,测得牙型角的侧边缘。分析螺纹图像边界与实际轮廓的关系,基于螺纹牙型与图像阴影区域边界差值方程校正测量结果。采用标定算法标定相机,得到相机径向畸变系数和内参数矩阵。螺纹牙型角检测实验结果表明,系统测得的牙型角与实际工具测量牙型角均值相差0.022°,可知,本系统能较准确地测量出螺纹牙型角,因而可以应用于螺纹牙型角的准确测量。To solve the problems of the low automation and efficiency of detection in the measuring process of included angle,an included angle measurement device is designed and motion control algorithm with image processing algorithm is developed based on machine vision.An industrial camera is used to obtain thread images,the Otsu algorithm with edge enhancement algorithm is used to obtain segmented images of thread,and then the LSD algorithm based on segmented image algorithm is used to measure the inclusive included angle.In addition,by analyzing the relationship between thread image boundary and actual thread profile,measurement results are corrected by the difference equation between thread profile and shadow image area boundary.The camera radial distortion coefficient and internal parameter matrix are obtained by camera calibration with a calibration algorithm,and some experiments about thread detection are done.The mean difference between the included angle results measured by this device and tool microscope is approximately 0.022°,which is more accurately,so this device can be applied to the accurate measurement of thread included angle.

关 键 词：机器视觉 牙型角测量 改进Otsu算法 LSD算法 

分 类 号：TG85[金属学及工艺—公差测量技术] TH74[机械工程—光学工程]