机器视觉在机械表面加工缺陷诊断中的研究  

Research on Machine Vision in Diagnosis of Mechanical Surface Processing Defects

作　　者：张涛[1] ZHANG Tao(Technician College(Chengdu Vocational and Technical College of Industry and Trade),Chengdu,Sichuan 611730,)

机构地区：[1]成都市技师学院(成都工贸职业技术学院),四川成都611730

出　　处：《自动化应用》2024年第21期40-42,47,共4页Automation Application

摘　　要：研究了基于机器视觉的机械表面缺陷检测方法,特别是应用MobileNet实现了缺陷识别。首先,介绍了机械表面缺陷检测系统的基本原理;其次,详细阐述了图像预处理和自适应阈值分割的数学原理;最后,构建并训练了基于MobileNet的缺陷检测模型,并在MATLAB环境中使用东北大学条带表面缺陷数据集进行测试。结果表明,该方法在准确率、精确率和召回率等指标上均表现优异,验证了其在复杂表面缺陷检测中的有效性和实用性。This paper investigates a machine vision based method for detecting mechanical surface defects,particularly the application of MobileNet for defect recognition.Firstly,this paper introduces the basic principle of the mechanical surface defect detection system.Secondly,the mathematical principles of image preprocessing and adaptive threshold segmentation are elaborated in detail.Finally,a defect detection model based on MobileNet is constructed and trained,and tested using the Northeastern University strip surface defect dataset in MATLAB environment.The results show that the method performs excellently in accuracy,precision,and recall,verifying its effectiveness and practicality in complex surface defect detection.

关 键 词：机器视觉 MobileNet模型 自适应阈值分割 缺陷检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]