基于二次全局阈值图像分割算法的产品表面缺陷自动化检测装置  

Automatic Detection Device for Product Surface Defects Based on Quadratic Global Threshold Image Segmentation Algorithm

作　　者：郭志龙 赵辉 董云 王杰 惠子文 Guo Zhilong;Zhao Hui;Dong Yun;Wang Jie;Hui Ziwen(Lanzhou Petrochemical University of Vocational Technology,Lanzhou Gansu 730060,China)

机构地区：[1]兰州石化职业技术大学,甘肃兰州730060

出　　处：《机械管理开发》2025年第1期131-134,共4页Mechanical Management and Development

基　　金：2024年高校教师创新基金项目“CRH型高速动车组转向架轴箱节点全自动化装配装置”(2024A-236);2024年教学研究项目“电气工程及自动化专业职业本科学生创新能力提升研究”。

摘　　要：开发了一套基于二次全局阈值图像分割算法的产品表面缺陷自动化检测装置。通过该装置的硬件设备采集图像,将图像进行灰度化、图像增强处理以及ROI选取等前期的预处理,选择最优的HALCON视觉系统参数并进行匹配检测,在该装置的HALCON视觉系统的HDevelop编译环境下开发基于二次全局阈值的图像分割算法,编写HALCON源码,通过该算法标记出所需的缺陷对象或特征的轮廓,提取出划痕的亚像素轮廓,根据面积和长度等特征来判断其是否为缺陷产品。将CRH380A型动车组转向架车轮擦伤的样本进行图像检测测试,取得了良好的缺陷检测效果。A product surface defect automated detection device based on a quadratic global threshold image segmentation algorithm has been developed.Specifically,images are collected through the hardware equipment of the device,and pre-processing such as grayscale,image enhancement,and ROI selection are performed to select the optimal HALCON visual system parameters and perform matching detection.In the HDevelop compilation environment of the HALCON visual system of the device,an image segmentation algorithm based on quadratic global threshold is developed,and the HALCON source code is written to mark the contours of the required defect objects or features through this algorithm,Extract sub-pixel contours of scratches and determine whether they are defective products based on features such as area and length.The image detection of samples of wheel scratches on the CRH380A high-speed train bogie achieved good defect detection results.

关 键 词：CRH380A型动车组 HALCON视觉系统 产品表面缺陷 预处理 二次全局阈值 划痕 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]