Enhanced land use/cover classification using support vector machines and fuzzy k-means clustering algorithms

Tao He,Chang-Ru Hu,Ji-De Xu,Xue-Jun Wang,Yu-Jun Sun

doi:10.1117/1.jrs.8.083636

Abstract

Land use/cover (LUC) classification plays an important role in remote sensing and land change science. Because of the complexity of ground covers, LUC classification is still regarded as a difficult task. This study proposed a fusion algorithm, which uses support vector machines (SVM) and fuzzy k-means (FKM) clustering algorithms. The main scheme was divided into two steps. First, a clustering map was obtained from the original remote sensing image using FKM; simultaneously, a normalized difference vegetation index layer was extracted from the original image. Then, the classification map was generated by using an SVM classifier. Three different classification algorithms were compared, tested, and verified—parametric (maximum likelihood), nonparametric (SVM), and hybrid (unsupervised-supervised, fusion of SVM and FKM) classifiers, respectively. The proposed algorithm obtained the highest overall accuracy in our experiments.

Highlights

Land use/cover (LUC) classification is a key research field in remote sensing, and plays an important role in climate change, biodiversity conservation, and people’s livelihoods
Many other advanced classification techniques have been introduced in the field of remote sensing classification, including artificial neural networks, machine-learning, decision trees, genetic algorithms, and support vector machines (SVM).[6,7,8,9,10]
In order to use both advantages of SVM and fuzzy k-means (FKM) clustering, we proposed a combination method to deal with LUC classifications in remote sensing images

Summary

Introduction

Land use/cover (LUC) classification is a key research field in remote sensing, and plays an important role in climate change, biodiversity conservation, and people’s livelihoods. Accurate LUC maps derived from remotely sensed data have become the basis for analyzing many socio-ecological issues.[1] LUC classification is nothing more than a convenient abstraction and may be improved by considering the other lines of evidence, such as surfaces that reflect the range of variability within and between the categories of a classification scheme.[2]. One basic issue to enhance the LUC classification is to choose an optimal classifier. A series of conventional classification methods have been well developed and long used for remote sensing applications, which are parallelepiped, minimum distance, and maximum likelihood (ML) models.[3,4,5] Many other advanced classification techniques have been introduced in the field of remote sensing classification, including artificial neural networks, machine-learning, decision trees, genetic algorithms, and support vector machines (SVM).[6,7,8,9,10]

Methods

Results

Conclusion