基于机器视觉的接线端子缺陷检测算法  

Defect detection algorithm of terminal blocks based on machine vision

作　　者：赵久强 曾豪辉 冯毅萍[1] 曹峥[1] 仲玉芳[1] ZHAO Jiuqiang;ZENG Haohui;FENG Yiping;CAO Zheng;ZHONG Yufang(College of Control Science and Engineering,Zhejiang University,Hangzhou 310027,China)

机构地区：[1]浙江大学控制科学与工程学院,浙江杭州310027

出　　处：《实验技术与管理》2022年第11期40-45,共6页Experimental Technology and Management

基　　金：2021年浙江大学实验技术研究项目(SYB202119);教育部产学合作协同育人项目(201902262041)。

摘　　要：接线端子是工业生产中常用的电路连接元件,决定了电子系统的安全性与可靠性。以常用的弹簧式直插接线端子为例,对其缺陷进行定义与分类,针对性地设计了其黄铜导电条与高强度弹簧的缺陷检测算法。首先对图像进行预处理校正,设计了定位基准拟合与位姿校正算法,对其形态归一化,之后分别设计基于四叉树分形的区域面积匹配与基于随机森林的细微特征检测,实现不同类别缺陷的检测。实验结果表明:对接线端子中导电条部分的检测正确率基本达到100%,而对高强度弹簧处缺陷准确率达98%以上,且对噪声具有一定鲁棒性,能够满足实际工业生产需求。Terminal blocks are commonly used as circuit connection components in industrial production, which determines the safety and reliability of electronic system. Taking the commonly used spring-type straight plug terminals as an example, the defects are defined and classified, and the defect detection algorithm for its brass conductive strips and high-strength springs is designed in a targeted manner. Firstly, the image is preprocessed and corrected, and the positioning reference fitting and pose correction algorithms are designed to normalize the morphology. Then the area matching based on quadtree fractal and the fine feature detection based on random forest are designed respectively, to achieve the detection of different categories of defects. The experimental results show that the detection accuracy of the conductive strip in the terminal is almost 100%, and the accuracy of the defect in the high-strength spring is more than 98%. The detection accuracy is robust to noise, which can meet the needs of actual industrial production.

关 键 词：接线端子 缺陷检测 四叉树图像分形 随机森林 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]