利用图像处理技术测量丝杠螺距的机器视觉系统  

A machine vision system for measuring screw pitch with image processing techniques

作　　者：韩天雨 路长厚[1] 李建美[1] 尹昂 侯秋林 HAN Tianyu;LU Changhou;LI Jianmei;YIN Ang;HOU Qiulin(School of Mechanical Engineering,Shandong University,Jinan 250061,Shandong,China)

机构地区：[1]山东大学机械工程学院,山东济南250061

出　　处：《山东大学学报（工学版）》2022年第3期80-85,共6页Journal of Shandong University（Engineering Science）

基　　金：国家自然科学基金资助项目(51975337)。

摘　　要：为解决接触式测量方式损伤工件以及测量设备造价昂贵的问题,设计一种基于机器视觉的低成本、高精度的丝杠螺距测量系统。该系统依靠MATLAB平台搭建,由图像采集系统、机械系统、步进控制系统、图像和数据处理4个子系统构成。利用亚像素图像拼接算法拼接丝杠子图像从而获得完整的丝杠图像,通过图像增强、降噪等预处理方法改善图像质量,采用形态学边缘提取和亚像素边缘定位的组合算法精确提取丝杠边缘,并由数据处理程序自动测量螺距误差。试验结果表明,机器视觉系统和标准仪器(JCS-040A)对3种型号滚珠丝杠螺距测量结果的相对误差分别不超过0.04%、0.06%、0.09%,该方法应用于丝杠工业测量中的可行性得以验证。In order to solve problems of damage to workpiece and high-cost of measuring equipment by contact measuring method,a low-cost and high-precision screw pitch measuring system based on machine vision was designed.The system was built on MATLAB and consisted of four sub-systems:image acquisition system,mechanical system,stepping control system,image and data processing system.Sub-pixel mosaic algorithm was used to splice the sub-images of the lead screw and a complete lead screw image could be obtained.Image quality was improved by image enhancement and filtering.Morphological edge extraction and sub-pixel edge location algorithm were used to extract the screw edge accurately,and pitch errors were measured automatically by data processing program.The experimental results showed that the relative error of the machine vision system and standard instrument(JCS-040A)for the measurement results of the three types of ball screw pitches didn̓t exceed 0.04%、0.06%、0.09%,respectively.The feasibility of the system in industrial measurement of lead screw was validated.

关 键 词：丝杠 机器视觉 螺距误差 图像拼接 图像处理 

分 类 号：TH136[机械工程—机械制造及自动化] TP139[自动化与计算机技术—控制理论与控制工程]