Feature Extraction and Recognition Based on the Biological Analysis

Key Words:Stored Product Pests; Biological Analysis; Feature Extraction; BP Neural Networks; Support Vector Machines; Recognition and Classification

Abstract:In this paper the extracted features including rectangularity, elongation, invariant moments and the four ratios of the stored product pests, which are the ratio of antennae area to torso area, the ratio of antennae perimeter to torso perimeter, the ratio of head and chest area to abdominal area, the ratio of head and chest perimeter to abdominal perimeter. Then these 13 characteristic parameters are input to BP neural network and SVM for recognition and classification. Form the results we can see that the 13 features in this paper can be well reflected the stable characteristic information of the stored product pests.

Volume:513-517

Translation or Not:no