THE EXPERT SYSTEM FOR IDENTIFICATION OF TORTRICINAE (LEPIDOPTERA) USING IMAGE ANALYSIS OF VENATION
作者:
Jingdong Liu,
期刊:
Insect Science
(WILEY Available online 1996)
卷期:
Volume 3,
issue 1
页码: 1-8
ISSN:1672-9609
年代: 1996
DOI:10.1111/j.1744-7917.1996.tb00394.x
出版商: Blackwell Publishing Ltd
关键词: computer;expert system;identification;image analysis;Tortricinae;venation
数据来源: WILEY
摘要:
AbstractIn the present research, the venation of 53 species of Tortricinae (Lepidoptera) known from Japan was analyzed using image analysis technique. A method to use the new character “vein points” for species identification was established, and a corresponding expert system was developed. In this method, the X and Y pixel coordinates of 22 “vein points” of the forewing of Tortricinae were measured and then used as a main character. By statistical analysis, the “vein points” were proved to be able to reflect the venation pattern stably and accurately. These “vein points” were input into a personal computer. Let the computer measure the X and Y coordinates of each “vein point” automatically, and calculate the dissimilarity among 22 “vein points” of the specimen and those of the known Operational Taxonomic Units (OTUs), and display the names of the known OTUs in order of increasing dissimilarity. As a result of simulation tests on the expert system, 98. 2 percent of the correct answers ranked Ist, the remaining 1.8 percent ranked 2nd and 3rd. In order to reach a 100 percentage of correct answers, other image and literal information of the 53 species were also input into the identification system as references. These informations include morphological figures of adult Tortricids, genitalia of males and females, literal description about morphology, habitat, distribution, host and so on. Using this expert system, people other than taxonomists can also identify specimen of Tortricinae to species accurat
点击下载:
PDF
(769KB)
返 回