蛾翅数学形态特征用于夜蛾分类和鉴定的可行性研究

摘要: 为探讨蛾翅数学形态特征(MMC)在夜蛾科分类鉴定中的可行性， 本文利用数字化技术获得和处理昆虫图像， 对鳞翅目夜蛾科6种夜蛾的右前翅提取矩形度、 延长度、 叶状性、 偏心率、 球状性、 似圆度和不变矩Hu1、 Hu2等13项与大小尺度和方向均无关的数学形态特征， 并利用方差分析、 逐步判别分析和聚类分析等方法研究了各项数学形态特征在昆虫分类上作为分类特征的可行性、 可靠性和重要性， 并且从数学形态学角度对夜蛾科6个种的亲缘关系进行了分析。分析结果认为矩形度和延长度2个形态特征对这6种夜蛾的分类鉴定没有显著意义， 从而筛选出11个形态特征作为分类变量， 它们的作用大小依次为： （偏心率、 Hu5、 Hu7）＞Hu2＞似圆度＞球状性＞Hu3＞（叶状性、 Hu1、 Hu6）＞Hu4。利用蛾翅的这些特征参数成功地实现了对夜蛾科6种夜蛾的分类鉴定， 基于这些特征参数的6种夜蛾的亲缘关系远近与基于传统形态学的系统进化观点相同。研究表明蛾翅数学形态特征可应用于蛾类昆虫的快速鉴定， 为未来逐步实现蛾类昆虫的自动识别奠定了基础。

Feasibility of math-morphological characters of wings for classification and identification of noctuid moths (Lepidoptera: Noctuidae)

To attest the feasibility of math-morphological characters(MMCs) of moth wings for classification and identification of Noctuidae,we used digital technology to acquire and process insect images.Thirteen MMCs,such as rectangularity,elongation,lobation,eccentricity,sphericity,roundness and movement invariants including Hu1 and Hu2,which are invariants to the image size and direction,were selected.The characters from right forewings of moths from six noctuid species were extracted in this study.The methods including variance analysis,stepwise discriminant analysis and cluster analysis were used to study the feasibility,reliability and importance of each MMC as classification attribute of insect classification.The kinship among the six noctuids was analyzed from the perspective of mathematical morphology.The analysis results showed that rectangularity and elongation had no significant contribution to classification and identification of the six noctuids.Therefore,11 morphological characters were selected as the classification variables.The contribution of these variables could be ranked as follows:(eccentricity,Hu5,Hu7)>Hu2>roundness>sphericity>Hu3>(lobation,Hu1,Hu6)>Hu4.Classification and discrimination of the six noctuids were realized by using the character parameters of moth wings.The phylogenetic relationship of the six noctuids based on the character parameters was consistent with that based on the traditional morphological observation.The results indicate that MMCs of moth wings can be used to identify moths rapidly and will be useful in gradually realizing the automatic recognition of moths in the future.