Feature Selection for Chili Leaf Disease Identification Using GLCM Algorithm

Abstract

In agriculture, detecting and diagnosing leaf diseases are a major concern. Tracking crop fields and identification of symptoms of the disease is important for farmers. Image processing is an aid to the identification and classification of leaf diseases. For leaf disease identification, there are three image features, i.e., texture, color, and shape. Texture features are more important elements of them. The feature selection process is critical to get the best accuracy and minimum time measurement. There are 2500 samples of chili leaves with five diseases in this analysis are train, and 1000 samples are gathered in the research dataset are a test. In this job, using the GLCM algorithm, two classifiers analyze texture features. The extracted texture features and target value are given during training as an input to the SVM and KNN classifier. The texture features. Contrast, energy, correlation, entropy, cluster_shade, cluster_provience, kurtosis, skewness are used for disease identification, respectively. We are labeled as Cercospora leaf spot, chili mosaic, powdery mildew, leaf curl, and healthy leaf in five groups. SVM provides 87.04% accuracy, and KNN provides 94.04% accuracy using the k-fold method