粗糙集模糊聚类分析法在昆虫分类研究中的应用.pdf

昆虫学报ActaEntomologicaSinica,February2006,49(1):106-111ISSN045426296

粗糙集模糊聚类分析法在昆虫分类研究中的应用

杜瑞卿1,张征田1,刘广亮2,武福华1
(11南阳师范学院生命科学系,河南南阳473061;21南阳师范学院计算机系,河南南阳473061)

摘要:本文根据昆虫图像,对半翅目、鳞翅目、鞘翅目的28种昆虫提取的形状参数、叶状性、球状性等7项数
学形态特征进行了粗糙集模糊聚类分析。在粗糙集处理的基础上,分别进行7指标和3指标(相对约简)两种不
同的模糊聚类分析法相比较。结果显示,在作为目级阶元分类指标时,各项特征的重要性依次为:(似圆度、偏
心率)>(亮斑数、球状性、圆形性)>(叶状性、形状参数);粗糙集分类正确率优于模糊聚类分析法;粗糙
集处理后的3指标分类正确率优于未处理的7指标分类正确率。结论认为,粗糙集理论在昆虫依据数学形态特征
进行分类方面与统计分析方法相比更有优势,粗糙集滤过指标后再进行模糊聚类法分析在昆虫分类研究上具有
重要意义。
关键词:昆虫分类;粗糙集;模糊聚类;数学形态特征
中图分类号:Q966文献标识码:A文章编号:045426296(2006)0120106206
Applicationofrough2settheoryandfuzzyclusteringanalysisininsect
taxonomy
DURui2Qing1,ZHANGZheng2Tian1,LIUGuang2Liang2,WUFu2Hua1(11DepartmentofBiology,Nanyang
NormalUniversity,Nanyang,Henan473061,China;21DepartmentofComputerScience,NanyangNormal
University,Nanyang,Henan473061,China)
Abstract:Basedon7math2morphologicalfeatures(MMFs),suchasformparameter,lobation,sphericity,
etc.extractedfromtheimagesof28speciesofinsectsoftheHemiptera,LepidopteraandColeoptera,the
applicationofrough2settheoryandfuzzyclusteringanalysisininsecttaxonomywasevaluated.Then,basedon
thedatapreparedwithrough2settheoryanalysis,fuzzyclusteringanalysiswith7indexesor3indexeswasmade
separatelytoassesstheirefficiency.Theresultsshowedthatwhenusedasindexesintaxonomyattheorder
level,theMMFswererankedinthefollowingorderaccordingtotheirimportance:(holenumber,sphericity,
circularity)>(roundness,eccentricity)>(lobation,shape2parameter).Theclassificationcorrectness
basedonrough2settheoryishigherthanthatbasedonfuzzyclusteringanalysis;andthecorrectnessoffuzzy
clusteringanalysiswith3indexesbasedrough2settheoryisalsohigherthanthatwith7indexes.Evaluatedby
theirapplicationininsecttaxonomy,therough2settheoryismoreefficientcomparedwithstatisticalanalysis
method.Themethodoffuzzyclusteringanalysiswiththeindexfiltratedbyrough2settheoryhashigh
applicationprospectininsecttaxonomy.
Keywords:Insecttaxonomy;rough2settheory;fuzzyclusteringanalysis;math2morphologicalfeatures

上世纪50年代末期,Sneath和Miche