| 427 | 5 | 21 |
| 下载次数 | 被引频次 | 阅读次数 |
为解决雪茄烟烟支外观质量检测自动化程度低、人为影响因素大等问题，研制了一种雪茄烟烟支外观质量检测装置，并构建了基于机器视觉技术和深度学习的烟支外观质量缺陷检测模型。通过检测装置采集雪茄烟柱面缺陷图像，并利用随机平移、翻转等方法对图像进行均衡和增强，建立数据集；使用K-means聚类算法生成适用于柱面缺陷数据集的先验框；对YOLOv5模型进行训练，并验证检测效果；采用大津法求得最佳阈值对雪茄烟端面空头图像进行分割，通过统计烟丝缺失比例实现空头缺陷检测。结果表明：(1)采用YOLOv5模型对柱面缺陷进行检测，多类别平均准确率MAP(Mean Average Precision)为87.7%，单张图像检测时间为13.1 ms/张，模型抗干扰能力强，检测精度和时间均优于对比模型。(2)YOLOv5模型能够准确识别和定位多尺寸、多种类以及密集缺陷，具有较强泛化能力和鲁棒性。(3)在不同亮度光照环境下，均能够实现空头缺陷检测。该方法可为提高雪茄烟产品质量提供支持。
Abstract：For the automation of cigar appearance inspection, a cigar appearance inspection device was developed, and a cigar appearance inspection model based on machine vision technology and deep learning was established. The images of defects of the cylindrical surface of cigars were captured by a test device. The captured images were equilibrated and augmented by random panning and flipping to create a data set. The K-means clustering algorithm was used to generate appropriate anchors for the cigar defect dataset. The YOLOv5 model was trained and verified. The optimal threshold was obtained using the OTSU method to segment the images of loose end cigars. The loose end defect was identified by statistically computing the proportion of tobacco loss from the cigar end. The results showed that: 1) Using the YOLOv5 model to identify the defects of cylindrical surface,the mean average precision(MAP)reached 87.7% and the time for going through a single image was13.1 ms. The YOLOv5 model featured strong anti-interference and advantages over other models in terms of inspection accuracy and efficiency. 2) The YOLOv5 model was able to accurately identify and locate multi-size, multi-category and densely distributed defects with strong generalization ability and robustness. 3) The loose end defect could be identified under different luminance lighting environments. This method supports the promotion of cigar quality.
[1]余邦林，徐恒，姬凌波，等.国产雪茄烟挥发性成分质量分数和组成差异[J].烟草科技，2023,56(3):61-69.YU Banglin,XU Heng,JI Lingbo,et al.Differences in mass fractions and compositions of volatile components in domestic cigars[J].Tobacco Science&Technology,2023,56(3):61-69.
[2]李捷，陆海华，王翔，等.基于机器视觉的烟支接装质量在线检测系统[J].烟草科技，2019,52(9):109-114.LI Jie,LU Haihua,WANG Xiang,et al.Online cigarette appearance inspection system based on machine vision[J].Tobacco Science&Technology,2019,52(9):109-114.
[3]张超凡，董浩，刘勇，等.基于视觉形态特征检测的烟梗切丝质量分析[J].烟草科技，2017,50(1):67-73.ZHANG Chaofan,DONG Hao,LIU Yong,et al.Analysis of stem cutting quality based on visual morphological feature detection[J].Tobacco Science&Technology,2017,50(1):67-73.
[4]袁国武，刘建成，刘鸿瑜，等.基于Res Ne St的烟支外观缺陷分类[J].云南大学学报（自然科学版），2022,44(3):464-470.YUAN Guowu,LIU Jiancheng,LIU Hongyu,et al.Classification of cigarette appearance defects based on Res Ne St[J].Journal of Yunnan University (Natural Sciences),2022,44(3):464-470.
[5]徐龙泉，王澍，董浩，等.面向卷烟爆珠放行检验的气泡缺陷检测方法[J].烟草科技，2020,53(10):96-102.XU Longquan,WANG Shu,DONG Hao,et al.Method for release inspection of cigarette capsules with bubble defects[J].Tobacco Science&Technology,2020,53(10):96-102.
[6]丛明，卢长奇，刘冬，等.基于Refine-ACTDD的铸件外观微小缺陷检测方法[J].计算机集成制造系统，2022,28(9):2815-2824.CONG Ming,LU Changqi,LIU Dong,et al.Detection method for tiny defects in casting appearance based on Refine-ACTDD[J].Computer Integrated Manufacturing Systems,2022,28(9):2815-2824.
[7]孙浩巍，张轲，龙杰，等.云南雪茄烟叶外观质量评价体系研究[J].现代农业科技，2020(14):211-214.SUN Haowei,ZHANG Ke,LONG Jie,et al.Study on evaluating system of appearance quality of Yunnan cigar tobacco leaves[J].Modern Agricultural Science and Technology,2020(14):211-214.
[8]李孝钊.基于深度学习的轻量化网络火灾烟雾检测方法研究[D].重庆：重庆理工大学，2022.LI Xiaozhao.Research on lightweight network fire and smoke detection method based on deep learning[D].Chongqing:Chongqing University of Technology,2022.
[9]王勋，毛华敏，李唐兵，等.基于迁移学习和R-FCN的电力设备红外图像识别算法[J].传感器与微系统，2021,40(1):147-150.WANG Xun,MAO Huamin,LI Tangbing,et al.Recognition algorithm for infrared image of power equipment based on transfer learning and R-FCN[J].Transducer and Microsystem Technologies,2021,40(1):147-150.
[10]汪睿琪，张炳辉，顾钢，等.基于YOLOv5的鲜烟叶成熟度识别模型研究[J].中国烟草学报，2023,29(2):46-55.WANG Ruiqi,ZHANG Binghui,GU Gang,et al.Recognition model of tobacco fresh leaf maturity based on YOLOv5[J].Acta Tabacaria Sinica,2023,29(2):46-55.
[11]陆振山.基于Res Net的烟叶成熟度判定研究与应用[D].桂林：桂林电子科技大学，2022.LU Zhenshan.Research and application of tobacco maturity determination based on Res Net[D].Guilin:Guilin University of Electronic Technology,2022.
[12]刘延鑫，王俊峰，杜传印，等.基于YOLOv3的多类烟草叶部病害检测研究[J].中国烟草科学，2022,43(2):94-100.LIU Yanxin,WANG Junfeng,DU Chuanyin,et al.Detection of various tobacco leaf diseases based on YOLOv3[J].Chinese Tobacco Science,2022,43(2):94-100.
[13]齐玥程，王燕，李丽，等.基于深度学习的在线烟叶等级判定研究[J].安徽农业科学，2023,51(3):235-239.QI Yuecheng,WANG Yan,LI Li,et al.Online tobacco leaf grade determination research based on deep learning[J].Journal of Anhui Agricultural Sciences,2023,51(3):235-239.
[14]Yao J,Qi J M,Zhang J,et al.A real-time detection algorithm for kiwifruit defects based on YOLOv5[J].Electronics,2021,10(14):1711.
[15]叶赵兵，段先华，赵楚.改进YOLOv3-SPP水下目标检测研究[J].计算机工程与应用，2023,59(6):231-240.YE Zhaobing,DUAN Xianhua,ZHAO Chu.Research on underwater target detection by improved YOLOv3-SPP[J].Computer Engineering and Applications,2023,59(6):231-240.
[16]郭兰申，李杨，黄凤荣，等.融合引导锚框算法的Faster-RCNN缺陷检测[J].机械设计与制造，2022,374(4):160-164.GUO Lanshen,LI Yang,HUANG Fengrong,et al.Faster-RCNN part defect detection based on guided anchoring algorithm[J].Machinery Design&Manufacture,2022,374(4):160-164.
[17]高洁，王战红，刘纲.基于FSSD的微光烟雾检测方法[J].电子测量技术，2021,44(5):123-128.GAO Jie,WANG Zhanhong,LIU Gang.Low light level smoke detection method based on FSSD[J].Electronic Measurement Technology,2021,44(5):123-128.
基本信息:
DOI：10.16135/j.issn1002-0861.2023.0332
中图分类号:TS47
引用信息:
[1]张鹏飞,陈姣文,郭洪滨等.雪茄烟烟支外观质量检测方法[J].烟草科技,2024,57(01):91-97.DOI:10.16135/j.issn1002-0861.2023.0332.
基金信息:
国家烟草专卖局标准预研项目“卷烟盒条包装外观质量在线视觉检测设备性能指标检测及溯源方法研究”（国烟科[2023]35号）; 四川中烟工业有限责任公司科技项目“雪茄烟支外观质量视觉检测设备计量测试技术研究与应用”（2022510000340243）