A feature selection method using slime mould optimization algorithm in order to diagnose plant leaf diseases
Seyed Mohamad Javidan, Ahmad Banakar, Keyvan Asefpour Vakilian, Yiannis Ampatzidis
Abstract
One of the most influential factors in the reduction of agricultural products is diseases in plants. In the meantime, every year, apple trees suffer huge losses in terms of plant diseases, and the yield of their product decreases. Therefore, the study of the identification of apple leaf diseases is of great importance. The use of machine vision and machine learning algorithms have greatly helped pathologists in this field. Accurate identification of effective features such as color, texture, and shape in the diagnosis of plant diseases in the processing of images taken from the diseased plant plays an important role in the classification of disease groups. In this study, a new optimized method named slime mould optimization algorithm and SVM classifier were combined to diagnose three apple tree diseases, i.e., black spot, black rot or frog eye leaf spot, and cedar rust. The results obtained in this method provided 12 effective features out of 159 features extracted from disease images, and the accuracy of disease classification was 96.21%.