Extension Theory for Classification of the Stored-Grain Insects

Abstract

The design of the classifier is one of the important parts of the online detection system of the stored-grain insects based on the image recognition technology. The classification of the insects was of many image feature parameters, and the mixing degree among feature parameters of various species of the insects was large. The extension theory was proposed to be applied to the automatic classification of the insects. A method that constructed the matter element matrix of the insects was put forward based on the mean and variance of the image features. After calculating the correlation degrees between the insect to be recognized and the nine species of insects, the insect could be recognized by the maximum integrated correlation degree criterion. The experiment confirms that the recognition of the insects based on the extension theory is practical and feasible by the training and analyzing of the samples of the insects.