Area of Mangrove Forests Calculated by Color Image Segmentation using K-Means Clustering and Region Growing)

Abstract

The calculation of the area of mangrove forests by conventional methods requires much time and energy. In this study, a tool for calculating the area of mangrove forests in Southeast Sulawesi Province, Indonesia, using satellite imagery is developed on the basis of two segmentation methods, k-means clustering and region growing. We then compare those two methods to obtain the optimal method to calculate the area of mangrove forests. Before this research, there were no researchers who calculated the area of mangrove forests in Southeast Sulawesi using both methods. We constructed a calculation algorithm using Matlab, which includes different stages of digital image processing. The area of mangrove forests is calculated on the basis of the number of pixels with an area density of 900 m2/pixel. The accuracy of the two segmentation methods is compared for identical areas obtained by the National Institute of Aviation and Space in Indonesia (LAPAN), i.e., the area obtained by LAPAN is used as a reference in calculating the accuracy. The accuracy of the region growing segmentation method is 33.33%, whereas that by the k-means clustering segmentation method under optimum conditions is 59.26% in the application of 12 clusters.