Detection of Harmful Insects Based on Gray-Level Cooccurrence Matrix (GLCM) in Rural Areas Mehdi Ebadi

Abstract

There are many types of insects that affect agricultural fields. These harmful insects should be classified in a smart implementation for the rural fields. The main point to detect depends on their texture color. These textures are different from one insect to another. We propose a new hybrid method based on Gray Level Co-occurrence Matrix (GLCM) to detect the harmful insects in agricultural fields. The main idea shows that a tested image is composed of different texture regions of the insect and this will help to extract feature value. This paper consists of three steps: the first step extracts texture features using GLCM in four directions which are 0, 90, 180 and 270 degrees from the gray image. The second step trains the neural network depending on texture features in a large number of variety insect's images. The third step tests the unknown insect's image to classify it whether harmful or not. The purpose of this study helps rural farmers to detect the harmful insects and classify them to take care of their crops.