摘要
为准确识别卷烟爆珠生产中产生的拖尾缺陷并计量拖尾高度,提出了一种基于球体多角度外轮廓圆投影的拖尾缺陷检测方法。首先,通过自旋转机构和相机快速采集单颗爆珠10个不同角度的图像;其次,使用轮廓提取算法提取每张图像中爆珠外轮廓;然后,使用轮廓圆投影方法将外轮廓展开,检测每个角度对应的爆珠外轮廓上异常突起部分,并计算高度值;最后,综合10个不同角度爆珠的轮廓信息,对爆珠的外表面是否有拖尾缺陷进行判定。分别挑选出200颗拖尾爆珠和200颗合格爆珠样品进行测试,结果表明本文方法的拖尾缺陷识别率≥95%,误检率<1%,最小拖尾识别高度为0.05 mm;选取10批次各1万颗左右的样品进行爆珠放行通过检验,结果显示本文方法与人工灯检方法的偏差分布为[-0.08%,0.21%],能够满足爆珠抽样检验要求。该方法可为卷烟爆珠生产质量控制提供技术支持。
In order to accurately detect the defective tailed capsules and measure the tail’s height,a method based on multiple-angle contour projection of sphere was proposed.First,capture images of a single cigarette capsule at 10 different angles by a self-rotating mechanism coupled with a high speed camera.Next,extract the contour of the capsule in each image by a contour extraction algorithm.Then expand the contour by the contoured circular projection method and identify any protuberance from the contour corresponding to each angle and calculate its height.Finally,identify the tail defect,if any,on the capsule surface based on the contour information gathered from the 10 images of the capsule.The proposed method was tested with 200 qualified capsule samples and 200 selected tailed-capsule samples,the results showed that the recognition rate of the proposed method was≥95%and the false positive rate was<1%;the minimum recognizable tail height was 0.05 mm.Release inspection was conducted with capsules of 10 selected batches(roughly 10 thousand capsules per batch),the results indicated that the deviation between the proposed method and manual inspection ranged from-0.08%to 0.21%,which met the requirements of sampling inspection for cigarette capsules.This method provides a technical support for the quality control in cigarette capsule production.
作者
王澍
徐龙泉
董浩
彭黔荣
周明珠
张龙
罗光杰
吴晓松
李志刚
刘勇
WANG Shu;XU Longquan;DONG Hao;PENG Qianrong;ZHOU Mingzhu;ZHANG Long;LUO Guangjie;WU Xiaosong;LI Zhigang;LIU Yong(Hefei Institutes of Physical Science of CAS,Hefei 230031,China;University of Science and Technology of China,Hefei 230026,China;Technology Center,China Tobacco Guizhou Industrial Co.,Ltd.,Guiyang 550000,China;China National Tobacco Quality Supervision&Test Center,Zhengzhou 450001,China)
出处
《烟草科技》
EI
CAS
CSCD
北大核心
2021年第1期77-84,共8页
Tobacco Science & Technology
基金
贵州中烟工业有限责任公司科技项目“基于机器视觉的烟用胶囊缺陷检测系统研制”(GZZY/KJ/JS/2016 CY 023-1)
中国科学院科技服务网络计划区域重点项目“自主导航消毒机器人的研制及应用示范”(KFJ-STS-QYZD-097)。
关键词
卷烟爆珠
计算机视觉
拖尾缺陷
球体
轮廓圆投影
Cigarette capsule
Computer vision
Tail defect
Sphere
Circular projection of contour
