蚁群聚类分析算法在茶叶等级分类识别中的应用  

Application of Ant Colony Clustering Algorithm in Classification of Tea Level

作　　者：郑华军[1] 张宪[1] 乔欣[1] 

机构地区：[1]浙江工业大学特种装备制造与先进加工技术教育部/浙江省重点实验室,浙江杭州310014

出　　处：《轻工机械》2011年第5期90-93,101,共5页Light Industry Machinery

摘　　要：来自非原产地的龙井茶已经严重影响了原产地茶叶的信誉与销售。为了减弱这种影响,文章提出了一种蚁群聚类算法应用在茶叶等级分类识别上,因为相对于其他算法,蚁群聚类分析对未知分类的茶叶实行自动分类更有优势。为综合分析茶叶的特性,采集了3个等级的茶叶,每个级别有60组样品,然后提取每个样品的图像和光谱特征共16个参数,将180组样品先自动随机分类。最后利用蚁群聚类分析算法实现样品自动归类。结果发现,与原分组比较后,基于蚁群聚类分析算法的分类识别率达到了92.2%。这表明利用蚁群聚类分析对未知茶叶等级分类是可行的。The existence of fake tea from non-origin impacts on the credibility and sales of the origin Longing tea seriously. In order to weaken this impact, the paper proposed a technology using ant colony clustering algorithm in classification of tea level. Compared to other algorithms, ant colony clustering algorithm has a priority to classify tea level. To acquire and analyze the characteristics of tea comprehensively, collecting tea of 3 levels with sixty samples for each level, then the 16 parameters of the images and spectra from each sample were collected, random categorized the 180 samples, at last to classify the samples automatically with ant colony clustering algorithm. It was found that model based on the ant colony clustering algorithm had the best performance with accuracy of 92.2%. The result shows the method can be used in the identification of tea level. [ Ch ,5 fig. 3 tab. 11 ref. ]

关 键 词：茶叶分级 蚁群算法 聚类分析 等级分类 

分 类 号：TP274.3[自动化与计算机技术—检测技术与自动化装置]