Automated Pomegranate Leaf Disease Identification: An Image Processing Perspectives

Agriculture is a predominant occupation in India, and to date, automated plant disease detection has been applied primarily to crops such as paddy, wheat, and various vegetables (e.g., tomato, cucumber, and potato). In this work, a novel approach is proposed for the automated detection and labeling of pomegranate leaf diseases by employing Support Vector Machine (SVM) for classification and K-Means Clustering for image segmentation. The study focuses on four specific plant diseases—Bacterial Blight, Alternaria Alternata, Cercospora Leaf Spot, and Anthracnose—as well as the detection of healthy leaves when no affected areas are present. To enhance the visibility of the leaf images, contrast enhancement is performed as a preliminary step. Subsequently, K-Means Clustering is applied to segment the input image in the Lab color space. A Multiclass SVM is then utilized to classify the segmented regions based on identified affected portions of the leaf. Finally, the Region of Interest (ROI) is determined to calculate and display the total affected area, facilitating precise disease-type detection. Experimental results indicate that the proposed method achieves an accuracy of 0.96, a precision of 0.95455, a recall of 0.9333, an F1 score of 0.978, and a kappa statistic of 0.947, thereby demonstrating the efficacy of the proposed system for automated pomegranate leaf disease detection.