Abstract:

Based on 7 math-morphological features (MMFs), such as form parameter, lobation, sphericity, etc. extracted from the images of 28 species of insects of the Hemiptera, Lepidoptera and Coleoptera, the application of rough-set theory and fuzzy clustering analysis in insect taxonomy was evaluated. Then, based on the data prepared with rough-set theory analysis, fuzzy clustering analysis with 7 indexes or 3 indexes was made separately to assess their efficiency. The results showed that when used as indexes in taxonomy at the order level, the MMFs were ranked in the following order according to their importance: (hole number, sphericity, circularity)>(roundness, eccentricity)>(lobation, shape-parameter). The classification correctness based on roughset theory is higher than that based on fuzzy clustering analysis; and the correctness of fuzzy clustering analysis with 3 indexes based rough-set theory is also higher than that with 7 indexes. Evaluated by their application in insect taxonomy, the rough-set theory is more efficient compared with statistical analysis method. The method of fuzzy clustering analysis with the index filtrated by rough-set theory has high application prospect in insect taxonomy.

Key words: Insect taxonomy, rough-set theory, fuzzy clustering analysis, mathmorphological features