基于皱缩特性的烟叶聚类分析研究

Classification and Grouping of Tobacco Leaves Based on Crumpling Characteristics

为了研究不同原烟皱缩特性之间的差异,采用图像处理技术测定了不同原烟的皱缩率,并运用K-均值聚类分析方法从整体和分部位的角度对原烟进行分类。结果表明:(1)不同产地、品种、部位烟叶皱缩率存在极显著差异性,说明皱缩率是烟叶固有性质,可用皱缩率表征烟叶原料的皱缩特性。(2)通过K-均值聚类分析方法将全部烟叶样品和分部位烟叶样品分成两类时,类间样品均衡性和类间分组差异性均较好,产地特征更明显。(3)两种分类方式结果基本一致,第Ⅰ类烟叶产地主要包括湖南郴州、四川凉山、陕西宝鸡、河南郑州、云南保山、福建永安;第Ⅱ类烟叶产地主要包括贵州惠水、贵州盘县、贵州务川、广东梅州。

In order to investigate the differences between the crumpling characteristics of different raw tobacco,the crumpling rates of different raw tobacco were measured using image processing techniques,and K-means cluster analysis was applied to classify the raw tobacco from the overall and part-wise perspectives. The results showed that:( 1) the crumpling rates of different origins,varieties and parts of tobacco were significantly different,indicating that the crumpling rate is an inherent property of tobacco and can be used to characterise the crumpling characteristics of raw tobacco.( 2) When all tobacco samples and part samples were divided into two categories by K-means cluster analysis,the balance of samples between categories and the variability of groupings between categories were better,and the origin characteristics were more obvious.( 3) The results of the two classification methods are basically the same,with Class I tobacco origins mainly including Chenzhou in Hunan,Liangshan in Sichuan,Baoji in Shaanxi,Zhengzhou in Henan,Baoshan in Yunnan,and Yongan in Fujian;Class II tobacco origins mainly include Huishui in Guizhou,Panxian in Guizhou,Buchuan in Guizhou,and Meizhou in Guangdong.