湖北农业科学 ›› 2025, Vol. 64 ›› Issue (9): 195-201.doi: 10.14088/j.cnki.issn0439-8114.2025.09.031

• 信息工程 • 上一篇    下一篇

色度值替代外观特征快速准确鉴别初烤烟叶品种

陈大江1, 金楹2, 张芬芬2, 赵晓军2, 杨雅馨2, 刘正玲2, 周鹏1   

  1. 1.云南农业大学烟草学院,昆明 650201;
    2.云南省烟草公司昆明市公司,昆明 650051
  • 收稿日期:2025-02-11 出版日期:2025-09-25 发布日期:2025-10-28
  • 通讯作者: 刘正玲(1979-),女,云南昆明人,农艺师,主要从事烟叶评级研究,(电话)13888716256(电子信箱)542811538@qq.com;周 鹏(1980-),男,安徽淮南人,实验师,博士,主要从事烟草栽培及烟草化学研究,(电话)15060066568(电子信箱)zhoupeng@ynau.edu.cn。
  • 作者简介:陈大江(2000-),男,云南曲靖人,硕士,主要从事烟草栽培与生理生化研究,(电话)18108743572(电子信箱)3058450377@qq.com。
  • 基金资助:
    国家自然科学基金项目(32160517); 云南省烟草公司昆明市公司科技计划项目(昆烟办综〔2023〕18号KMYC202303)

Chroma values replacing appearance characteristics for rapid and accurate identification of flue-cured tobacco leaf varieties

CHEN Da-jiang1, JIN Ying2, ZHANG Fen-fen2, ZHAO Xiao-jun2, YANG Ya-xin2, LIU Zheng-ling2, ZHOU Peng1   

  1. 1. College of Tobacco, Yunnan Agricultural University, Kunming 650201, China;
    2. Kunming Branch of Yunnan Tobacco Company, Kunming 650051, China
  • Received:2025-02-11 Published:2025-09-25 Online:2025-10-28

摘要: 为优化现有初烤烟叶品种识别效率和准确度,进一步提升卷烟工业烟叶可用性。基于CIELAB颜色空间理论,利用色度仪对昆明产区3个品种初烤烟叶色度值进行测定,建立一种可以快速、准确识别烟叶品种的方法。结果表明,采用色度值与外观特征2种识别方法均可以实现不同品种初烤烟叶品种的识别,但色度仪测定的色度值更加准确可靠,色度值可较好地反映烟叶外观质量。不同样本虽有少数重叠,但总体聚类效果较好,表明同一部位不同品种色度值之间存在差异性,可以有效表征各品种的特征。采用明度值(L*)、红绿色度值(a*)、黄蓝色度值(b*)、饱和度(C*)及色差值(ΔE)等指标建立Fisher判别模型,模型交叉验证的结果显示,训练集的识别率为83.3%,交叉验证的识别率为81.1%,二者结果接近且均处于较高水平,表明该模型可以较好地识别初烤烟叶品种。该方法能够较好避免环境与人为因素对烟叶品种识别的干扰,推动不同品种初烤烟叶的鉴别从定性走向定量、从经验化转向理论化。

关键词: 色度值, 外观特征, 快速准确, 鉴别, Fisher判别模型, 初烤, 烟叶品种

Abstract: This paper aimed to optimize the efficiency and accuracy of existing identification methods for flue-cured tobacco leaf varieties and further enhance the usability of tobacco leaves in the cigarette industry.Based on the CIELAB color space theory, chroma values of three varieties of flue-cured tobacco leaves from the Kunming production area were measured using a chroma meter, and a method for rapid and accurate identification of tobacco leaf varieties was established. The results showed that both chroma values and appearance characteristics could be used to identify different varieties of flue-cured tobacco leaves, but the chroma values measured by the chroma meter were more accurate and reliable, and the chroma values could better reflect the appearance characteristics of tobacco leaves.Although there was a small amount of overlap between different samples, the overall clustering effect was good, indicating that there were differences in chroma values between different varieties from the same position, which could effectively characterize the features of each variety. A Fisher discriminant model was established using indicators such as lightness (L*), red-green chroma value (a*), yellow-blue chroma value (b*), saturation (C), and color difference (ΔE). The cross-validation results of the model showed that the recognition rate of the training set was 83.3%, and the cross-validation recognition rate was 81.1%. The two results were close and both at a high level, indicating that the model could effectively identify flue-cured tobacco leaf varieties. This method could better avoid the interference of environmental and human factors in the identification of tobacco leaf varieties, promoting the identification of different varieties of flue-cured tobacco leaves from qualitative to quantitative and from empirical to theoretical.

Key words: chroma values, appearance characteristics, rapid and accurate, identification, Fisher discriminant model, flue-cured, tobacco leaf varieties

中图分类号: