多级图像分类系统在硅钢冷轧表面缺陷检测中的应用  

Application of multi-level image classification system in surface defect detection of cold-rolled silicon strip

作　　者：甘胜丰[1] 孙林[1] 曹阳[2] 刘广益[2] 

机构地区：[1]中国地质大学(武汉)机械与电子信息工程学院,湖北武汉430074 [2]武汉钢铁(集团)公司

出　　处：《冶金自动化》2009年第2期63-65,共3页Metallurgical Industry Automation

基　　金：湖北省自然科学基金资助项目(2006ABA346)

摘　　要：针对硅钢冷轧生产过程中带钢表面缺陷的特点,采用一种基于专家经验和决策树分类器的多级带钢表面缺陷图像分类系统,包含训练集与测试集的采集、特征提取、专家经验训练分类器、决策树分类器等。实际应用表明,该系统能较好地识别结疤、重皮、孔洞和边裂等严重缺陷图像,提高了武钢硅钢冷轧生产过程的自动化水平。According to features of surface defect, on cold-rolled silicon strip, a multi-level image classification system for strip surface defect based on expert knowledge and decision tree classifier was adopted. The system includes collection of training set and test image, extraction of image feature, expert knowledge training classifier and decision tree classifier. Practical applications show that the system can recognize scab, double skin, holes and edge crack etc. serious defects and automatic level of cold-rolled silicon strip production process is improved.

关 键 词：冷轧硅钢带钢 表面缺陷检测 图像分类 专家经验 决策树 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]