摘要
在科分类阶元上对半翅目、鳞翅目和鞘翅目 8个科的 2 3种昆虫图像中提取的昆虫面积、周长等 11项数学形态特征进行了统计分析。结果表明 ,在科的阶元上 11项特征可靠性大小依次为 (似圆度、偏心率 ) >(面积、周长、横轴长、球状性 ) >(纵轴长、圆形性 ) >(形状参数、叶状性 ) >亮斑数。从数学形态学角度出发 ,夜蛾科等 3个科的亲缘关系远近为夜蛾科与粉蝶科 >大蚕蛾科与粉蝶科 >夜蛾科与大蚕蛾科 ;鳃金龟等 3科的亲缘关系远近为鳃金龟科与天牛科、丽金龟科与天牛科 >鳃金龟科与丽金龟科。
Statistical analysis of 11 math-morphological feature (MMF) (such as area, perimeter, etc.) from images of 23 species of insects of the Pentatomidae, Coreidae, Noctuidae, Saturniidae, Pieridae, Melolonthidae Rutelidae and Cerambycidae families indicates that the ranked reliability of MMF in the identification of insect families is, from high to low: (roundness, eccentricity)>(area, perimeter, X-length, sphericity)>(Y-length, circularity)>(form factor, lobation)>hole number. From the perspective of mathematical morphology, the kinship between the Nuctuidae, Saturniidae and Pieridae can be ranked as follows: Nuctuidae and Pieridae> Saturniidae and Pieridae> Nuctuidae and Saturniidae. Kinship between the Melolonthidae, Rutelidae and Cerambycidae can be ranked as follows: Melolonthidae and Cerambycidae> Rutelidae and Cerambycidae> Melolonthidae and Rutelidae.
出处
《昆虫学报》
CAS
CSCD
北大核心
2003年第3期339-344,共6页
Acta Entomologica Sinica
基金
国家"8 6 3"项目 ( 86 3- 30 6 - ZD0 5- 0 2- 0 3)
国家自然科学基金项目 ( 30 2 70 16 8)
关键词
数学形态特征
昆虫分类
计算机视觉技术
科阶元
math-morphological feature (MMF)
insect classification
computer vision technology
family
