The Design And Implemention On Insect's Wing Image Classfication Algorithm

With the in-depth research of computer vision technology, image processing and pattern recognition technology is widely used. It's a new exploration in the field of insect identification that using insect wings images classify species of insects. Currently, insect image processing requires higher demand having research areas in insect biology, agriculture, forestry, pest and disease control, and in the field of customs inspection and quarantine of imported products. Insect image recognition technology can realize the automatic identification of insects, and facilitate insect experts to classify insects quickly and more accurately, which is completely depending on the image characteristics. So, it is much less time-consuming and labor-exhaustive.In this thesis, we focus on insect wings image preprocessing, image feature extraction, feature matching selection criteria. Main contributions are:(1) It's proposed that insect wings image preprocessing method based on normalization of direction and scale. The main processes include color images of different resolution normalization which is providing a unified data for follow-up feature extraction.(2) In our designed method, we carry out image feature extraction algorithm. The main methods include image color features, texture features and SIFT features. (3) We establish the effective strategy for insect identification and build insect identification algorithm based on wings feature. Experimental result shows algorithm's availability.