Early Blight Disease Segmentation on Tomato Plant Using K-means Algorithm with Swarm Intelligence-based Algorithm

Early blight commonly attacks the tomato leaf. The drone and image processing technologies offer a tool to detect the symptom of the tomato disease. One of the stages in the early blight detection is the segmentation of the tomato leaf. The K-means algorithm is a known image segmentation method which works simply and quickly. However, randomized initialization of centroids causes the K-means algorithm to easily get stuck in the local optimum and, as a result, gives imprecise segmentation. The swarm intelligence-based algorithm can avoid this problem. For this reason, we propose the early blight disease segmentation method on tomato leaves uses K-means algorithm with swarm intelligence-based algorithm. This research uses the Particle Swarm Optimization (PSO) because PSO is one of the swarm intelligence-based algorithms which has a balance in exploration and exploitation. The Hue of the HSV color space is used as input. From the experimental results, we can conclude that the performance of the early blight disease segmentation method using the K-means algorithm with swarm intelligence-based algorithm is much better.