Implementing PSO-based Image Segmentation for Detecting Sweet Potato Leaf Disease

Abstract

Sweet potato (Ipomoea batatas) is an important global crop, but its production is threatened by various leaf diseases, requiring accurate and efficient disease detection methods. Traditional manual inspection is labor-intensive and error-prone, making automated image processing techniques a promising alternative. This study implements Particle Swarm Optimization (PSO)-based image segmentation to detect diseased leaf regions by optimizing threshold selection in HSV color space. In the classification phase, leaves are classified into healthy and diseased classes using a Euclidean distance-based classifier. The proposed method achieved an average classification accuracy of 88.1%, with an accuracy of 95.8% for diseased leaves and 80.4% for healthy leaves, demonstrating its effectiveness in discriminating infected regions. The results confirm that PSO is a robust and efficient segmentation technique that improves the accuracy of disease detection. This research highlights the potential of PSO-based segmentation in smart agriculture, enabling early disease detection to help farmers take timely action and minimize crop losses. Compared to traditional methods, PSO reduces computational complexity while maintaining high segmentation accuracy, making it a valuable tool for agricultural disease monitoring. Future work can integrate deep learning models to refine disease classification and expand datasets to improve system performance under different environmental conditions.