Assessment of Pleiades Satellite Image for Mangrove Family Classification

Abstract

Remote sensing technology is the most common method used in monitoring conservation and restoration at mangrove areas. This study aims to classify the mangrove family at Bagan Datuk, Perak, using object-based image analysis techniques based on Pleiades’ image with 0.63m spatial resolution obtained from the Malaysian Remote Sensing Agency (ARSM). The segmentation was done by choosing a suitable scale and merge level. Two classifiers namely support vector machine (SVM) and k-nearest neighbor (KNN) were used to classify the mangrove family. The mangrove family map was produced from the higher accuracy of the classification. The results show that the overall accuracy of SVM is 63.81% (kappa = 0.55) while KNN is 59.83% (kappa = 0.50). In conclusion, SVM outperformed K-NN for mangrove family classification.