HSO-Based K-Means Segmentation for Identification of Disease and Deficiency in Paddy Plant

In agriculture, image processing is frequently used to identify issues with fruit grading, weed identification, disease diagnosis, and other related tasks. In order to detect nutritional deficiencies and paddy plant disease, this study employs image processing. Image has to classified as diseased or nutrient deficient after performing segmentation of image and selection of important features from the image. There are many different image segmentation algorithms available. K-means is an incredibly easy algorithm to understand, apply, and provide accurate results with. Despite its accuracy and simplicity, its drawbacks include the requirement to estimate the value of K and choose K’s initial centroids at random from the provided data values. A unique image segmentation technique was developed to solve this problem. It is based on the identification of inadequacies using K-means and harmonic search optimization (HSO). The K-means algorithm’s performance is significantly impacted by the randomly chosen values for K and the centroid values of the clusters. The standard K-means method takes longer to produce accurate results because it needs random initialization. A K-means method based on HSO is suggested to expedite the initialization procedure. The aim function of the suggested approaches is the Otsu technique with an initial step created from the HSO. The HSO method is used to locate the cluster canters, which are then used to begin the K-means algorithm. Lastly, when segmentation results are compared to those obtained using the fuzzy C-means and classic K-means segmentation algorithms, the recommended HSO-based K-means technique performs better than the rivals.