Digital image processing technique for palm oil leaf disease detection using multiclass SVM classifier

Disease in oil palm sector is one of the major concerns cause it effects the production and economy losses to Malaysia. The problem of disease that arises in oil palm plantation are. Nowadays plant diseases detection has received a lot of attention in monitoring the symptoms at earlier stage of plant growth. This work presents the use of digital image processing technique for detection and classification of oil palm leaf disease symptoms. Here, the disease detection used k-means clustering and multiclass SVM classifier to determine two palm oil diseases based on the symptoms of the disease through its leaf. By using k-means clustering technique, thirteen types of features are extracted from the leaf images. The classification of the disease is carried out by using multiclass SVM classifier. The detection shows that SVM achieves accuracy of 97% for Chimaera and 95% for Anthracnose.