Improving feature extraction using Part Separating algorithm : Case study forinsect identification of Order Lepidoptera

摘要

In Automated Insect Identification (AII) research field, several researchesfocused on the feature in feature extraction that effective for insect classification in different taxonomic level. General taxonomic featuresfrom each taxa suchas color, texture and shape wereextracted and combined into asingle value from the whole insect body. This paper wasto develop feature extraction in AII for families identification of Order Lepidoptera by using Part Separating algorithm (PS), which generates a single feature value by separating feature from the whole insect body into five values. The fivefeatureswere extracted from head part, front-wing part, back-wing part, abdomen part and symmetry half,the extracted feature related with eachother. This proposed algorithm developed feature extraction method for family's identification at Order Lepidoptera. In result,proposed algorithm has accuracy 97.27 % at family Sphingidae.