Segmentation of rice leaves image for disease classification with K-means and GLCM

Agriculture as one of the industrial sectors is part of the work that is needed to support basic needs. Rice plants are plants that are very susceptible to pests, bacteria, or viruses. The introduction of the types of pests that attack is an important first step to support success in an agricultural sector. These pests of rice plants can be an obstacle for farmers to increase their yields, because these pests can damage crops, causing crop failure. Therefore, it is necessary to classify pests that attack rice plants, from the image of rice leaves this can be done. To find accuracy in the classification process, the GLCM and K-Means algorithm methods are used. This is done so that it can be one way to deal with pests properly, so that crop failure does not occur. This study uses the Rice Leaf Diseases dataset obtained from the UCI Machine Learning Repository website. This dataset contains 120 leaf image data consisting of three types of diseases including Bacterial Leaf Blight, Brown Spot and Leaf Smut. From the 120 data then divided into two parts, namely training data with a total of 100 and 20 testing data. The measurement of the accuracy value is calculated with the test sample with the correct result compared to the wrong one. The results of K-Means Clustering and GLCM segmentation resulted in an accuracy of 85.71% for the Bacterial leaf blight class, 86% Accuracy Values for the Brown Spot class, and 83.6% Accuracy for the Leaf Smut class. So the results of the highest accuracy of leaf image classification on rice plants are Brown Spot disease class/disease by producing 86% accuracy of the total sample used.