Classification And Recognition Based On Fuzzy Theory Warehouser Insects

Storedproducted insects, are the main objects in the economical insects research. The target what it encroaches on involves various kinds of goods and materials that human life needs. With our rapid economic development, the international trade is becoming more frequent which objectively offer advantageous condition for these storedproducted insects to propagate and spread. So strengthening the quarantine of storedproducted insects has important significance to raise the technological level of preventing and killing them, to protect stored material, to lesson loss and to speed up the national economic development.For many years, the insect's researchers have been making great efforts to look for a kind of scientific, fast, accurate, practical method for appraising the kind of storedproducted insects. In recent years, along with the expanding of computer technology applicated field, the technology of image processing and the methods of pattern recognition made a breakthrough. At present, utilizing computer technology to discern and classify storedproducted insects automatically is one of the important subject in insects' research field.This paper tries to classify and discern the storedproducted insects with the fuzzy set theory. The fuzzy set theory is put forward by the autocontrol expert of U.S.A. L. A. Zadeh professor in 1965. Although it is raised relatively late , its applications in each field are very actively at present. This paper introduces briefly the principle of the fuzzy set theory and explains one learning method without supervised in detail, and two kinds of the fuzzy patter recognition methods.First this paper does some preprocessing with enhancing image by median filtering and the histogram balancing technology. Then it uses computer to extract and unify 17 characteristics including the grey statistical feature, the texture feature and form feature from the preprocessed storedproducted insects' image automatically. Then it forms the standard patter storehouse by the improved fuzzy c-means algorithm, and designs classification machine by fuzzy patter recognition on the basis of approximation value and the principle of selecting the near. At last it puts the subsequence storedproducted insects into the classification machine and identifies them.With the help of pest research group in ZhengZhou institute of technology. I obtain nine pieces of specimen image-three kinds of storedproducted insects' in the civil as the study sample of classification machine and then utilize micro- camera of optics gather 30 pieces of storedproducted image. Using the methods mentioned above to discern, the discerning rate is always up to 88%. I feel quite satisfied with the analytic results of initial experimental data. The research of this project has wide application prospect and present a new approach to identify and classify storedproducted insects.