基于机器视觉的袋泡茶包缺陷检测方法  

Method based on machine vision for teabag defects detection

作　　者：杨庆华[1] 王玲[1] 荀一[1] 鲍官军[1] 张盛 

机构地区：[1]浙江工业大学特种装备制造与先进加工技术教育部重点实验室,浙江杭州310032 [2]浙江茶乾坤食品股份有限公司,浙江湖州313100

出　　处：《浙江工业大学学报》2015年第2期163-167,共5页Journal of Zhejiang University of Technology

基　　金：浙江省特种装备制造与先进加工技术重点实验室开放基金资助项目(2011EM002)

摘　　要：针对袋泡茶包的缺陷,设计了基于机器视觉的袋泡茶包缺陷检测装置,提出了识别袋泡茶包缺陷的图像处理方法.通过最大类间差算法对茶包图像进行阈值分割,去除背景.根据茶包外形特点以及不同的缺陷特征,对茶包进行四边形拟合并划分成不同区域.通过计算茶包边缘夹角以及统计分析各区域灰度信息的方法,识别斜包、皱包、空包、夹渣包、左右不对称包缺陷;通过形态学运算和边缘检测识别破包缺陷.实验结果表明:各种缺陷的正检率均达到85%以上,总的缺陷正检率达91.8%.To detect the defects of teabags, a defects detection device based on machine vision was proposed, and an image processing method was developed in this paper. The otsu algorithm was used to segment the object image so the background can be removed. According to the shape feature and different defects＇ characteristics of the teabags, rectangle fitting was applied to fit the teabag outline. As the included angles of the teabag edge lines were figured out, the inclined teabags can be found out. The teabag was divided into several areas within the rectangle. As the gray value information of each area was drawn, the teabag, whether it was a crinkle bag, empty bag, left-right asymmetry bag or had dross contained in margin, can be tested out. Morphological algorithm and border detection algorithm were operated to extract the damage defect of the teabag. The experiment verified that all the detection rates can be higher than 85%, and the overall detection rate is 91.8%.

关 键 词：机器视觉 最大类间差 区域划分 灰度信息 形态学 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]