| 摘要: |
| 用虫体面积、周长等11项数学形态特征对40种昆虫实现自动鉴别，得出了各项数学特征的权重。在昆虫的自动鉴别中，11项特征所起的作用大小为：面积＞偏心率＞形状参数＞周长＞纵轴长＞孔洞数＞横轴长，似圆度＞叶状性＞圆形性＞球状性。以面积、周长等数学形态特征为分类依据，对隶属8目25科的40种昆虫进行了二叉式分类，并以此为鉴别机理，实现了对昆虫自动识别软件Bug Visux的升级，使其能够自动鉴别的昆虫种类由3种增加到40种，准确率达到97．5％。 |
| 关键词: 数学 形态特征 应用 昆虫 自动鉴别 分类机理 自动识别软件 昆种品种 |
| DOI： |
| 修订日期：2001-08-30 |
| 基金项目:高等学校博士点专项科研基金，国家高技术研究发展计划课题资助项目 (863- 30 6- ZD0 5- 0 2 - 0 3) |
|
| On Computer-aided Insect Identification Through Math-Morphology Features |
|
|
|
| Abstract: |
| The paper presents the result of computer aided identification of 40 species of insects by using their 11 mathematical morphological features (MMF). The weight and importance of each MMF is evaluated on the basis of their reference frequency in the process of identification. The result demonstrated that the roles of 11 MMF in computer aided insect identification (CAII) is, from high to low: area>eccentricity>formfactor>perimeter> llength >Holenumber>(x length, roundness)>lobation>circularity>sphericity. According to 11 MMF, 40 species of insect which belonging to 8 Orders, 25 Families were identified through dichotomous method. And it also was the identification mechanism of CAII. Finally, the authors put forward an updated version of BugVisux and CAII software that can identify 3 species of insects, and thus enable it to identify as much as 40 species of insects with an accuracy rate as high as 97.5%. |
| Key words: MMF(mathematical morphological feature),CAII(computer aided insect identification),identification mechanism |
