基于贝叶斯网络的铜带表面缺陷图像分类  

Image Classification of Copper Strip Surface Defects Based on Bayesian Network

作　　者：赵鹤 杨晓洪[1] 李小彤 张果[1] ZHAO He;YANG Xiao-hong;LI Xiao-tong;ZHANG Guo(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650000,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650000

出　　处：《控制工程》2022年第10期1901-1906,共6页Control Engineering of China

基　　金：国家重点研发计划项目(2017YFB0306405);云南省重点研发计划项目(2018BA070);国家自然科学基金资助项目(61364008)。

摘　　要：铜带的表面质量是直接影响最终产品性能的参数之一,企业对铜带表面缺陷的有效检测能够及时发现问题并加以控制。稳定的缺陷图片获取系统和快速的分类模型都是在线缺陷检测的基础。首先,建立缺陷图像获取系统;然后,对缺陷图像进行去噪、增强处理,提取可以表达图像特征的信息,采用贝叶斯网络进行缺陷分类检测,并对3种贝叶斯网络进行了详细的评估和对比。结果表明,贝叶斯网络在铜带表面缺陷图像分类中具有稳定的网络结构和较好的缺陷识别效率,且计算结果稳定,运算时间短。The surface quality of the copper strip is one of the parameters that directly affect the performance of the final product.The effective detection of the surface defects of the copper strip can help enterprises to find and control the problems in time.A stable defect image acquisition system and a fast classification model are the basis of online defect detection.First,a defect image acquisition system is established in this paper.Then,the defect images are denoised and enhanced to extract the information that can express the image features.Bayesian networks are used to perform defect classification and detection,and three Bayesian networks are evaluated and compared in detail.The results show that the Bayesian network has a stable network structure and good defect recognition efficiency in the classification of copper strip surface defects,and the calculation results are stable and the calculation time is short.

关 键 词：带材表面质量 缺陷分类 贝叶斯网络 机器学习 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]