基于计算机视觉的卷烟包灰检测方法  

Methods for detecting ash integrity of burning cigarette based on computer vision

作　　者：董浩 钟宇[4] 王澍[2] 徐羽鹏 李晓辉[3] 张龙 周明珠[3] 邢军[3] 刘勇[1,2] 胡清源[1,3] DONG Hao;ZHONG Yu;WANG Shu;XU Yupeng;LI Xiaohui;ZHANG Long;ZHOU Mingzhu;XING Jun;LIU Yong;HU Qingyuan(University of Science and Technology of China,Hefei 230026,China;Hefei Institutes of Physical Science of CAS,Hefei 230031,China;China National Tobacco Quality Supervision and Test Center,Zhengzhou 450001,China;Hefei Design Institute of CNTC,Hefei 230041,China)

机构地区：[1]中国科学技术大学,合肥市230026 [2]中国科学院合肥物质科学研究院,合肥市230031 [3]国家烟草质量监督检验中心,郑州450001 [4]中国烟草总公司合肥设计院,合肥市230041

出　　处：《烟草科技》2024年第5期91-97,112,共8页Tobacco Science & Technology

基　　金：烟草行业标准项目“卷烟包灰性能测试方法”(2021B023);中国烟草总公司重大科技项目“烟草行业质量监控大数据构建及应用研究”[110202101080(SJ-04)]。

摘　　要：为解决卷烟包灰检测结果存在一致性和可比性差的问题,开展了基于计算机视觉的卷烟包灰检测方法标准化研究,提出了包灰颜色、裂口率、缩灰率、炭线宽度和整齐度等指标的检测方法和结果表征方式,研究了成像时机、检测区域、成像方式等因素对包灰检测结果的影响,通过共同实验评价了检测方法的重复性和再现性。结果表明:①通过包灰颜色、裂口率等指标能够全面评价卷烟包灰质量。②包灰检测可采用单面成像方式采集图像,应在卷烟阴燃状态时成像,检测区域的灰柱长度应大于25.0 mm。③基于计算机视觉的检测方法对于包灰颜色、缩灰率和炭线宽度具有较好重复性和再现性,裂口率和炭线整齐度重复性受样品影响较大。基于计算机视觉的包灰检测方法能够实现包灰质量的科学准确评价。To promote the consistency and comparability of the results in ash detection,the standardization of cigarette ash detection methods based on computer vision was studied,and the methods for determining ash color,ash crack rate,ash shrinking rate,char line width,and char line uniformity and their characterization were proposed.The effects of imaging timing,detection area,and imaging mode on the detection results were analyzed,and the repeatability and reproducibility of the proposed methods were evaluated through collaborative experiments.The results showed that:1)The ash integrity of cigarette could be comprehensively evaluated by the indexes including ash color,ash crack rate,etc.2)The image of the ash could be collected with single-sided imaging mode when the cigarette was smoldering,and the length of the ash column in the detection area should be longer than 25.0 mm.3)Computer vision-based methods showed good repeatability and reproducibility in ash color,ash shrinking rate,and char line width detection.The repeatability in ash crack rate and char line uniformity detection was susceptible to sample variation.The computer vision-based detection methods could scientifically and accurately evaluate the ash integrity of cigarettes.

关 键 词：卷烟 包灰颜色 裂口率 缩灰率 炭线宽度 炭线整齐度 计算机视觉 

分 类 号：TS439.2[农业科学—烟草工业]