Extracting Land Use/Cover of Mountainous Area from Remote Sensing Images Using Artificial Neural Network and Decision Tree Classifications: A Case Study of Meizhou, China

Abstract

Accurate land use/cover (LUC) classification data derived from remotely sensed data are very important for land use planning and environment sustainable development. Traditionally, statistical classifiers are often used to generate these data, but these classifiers rely on assumptions that may limit their utilities for many datasets. Conversely, artificial neural network (ANN) and decision tree (DT) classifications provide nonlinear means to extract LUC from remote sensing images without having to rely on statistical procedures or assumptions. This article used ANN and DT classifiers to extract LUC from remote sensing images which had been corrected with ancillary atmospheric and topographic data in the mountainous area of Meizhou and compared their accuracies with the statistical minimum distance (MD) classifier. Results show that the overall accuracies of LUC classifications are approximately 97.77%, 93.08% and 89.12%, and the kappa coefficients get to 0.97, 0.90 and 0.84 for the ANN, DT and MD methods, respectively, indicating that the ANN has a better accuracy than the DT and MD classifiers. It is suggested that ANN is a more effective method for remote sensing image classification of mountainous areas because of its higher accuracy and performance than DT and MD classifiers.