Improved Support Vector Machine Enabled Radial Basis Function and Linear Variants for Remote Sensing Image Classification.

Abdul Razaque,Muder Almi’Ani,Mohamed Ben Haj Frej,Bandar Alotaibi,Munif Alotaibi

doi:10.3390/s21134431

Abstract

Remote sensing technologies have been widely used in the contexts of land cover and land use. The image classification algorithms used in remote sensing are of paramount importance since the reliability of the result from remote sensing depends heavily on the classification accuracy. Parametric classifiers based on traditional statistics have successfully been used in remote sensing classification, but the accuracy is greatly impacted and rather constrained by the statistical distribution of the sensing data. To eliminate those constraints, new variants of support vector machine (SVM) are introduced. In this paper, we propose and implement land use classification based on improved SVM-enabled radial basis function (RBF) and SVM-Linear for image sensing. The proposed variants are applied for the cross-validation to determine how the optimization of parameters can affect the accuracy. The accuracy assessment includes both training and test sets, addressing the problems of overfitting and underfitting. Furthermore, it is not trivial to determine the generalization problem merely based on a training dataset. Thus, the improved SVM-RBF and SVM-Linear also demonstrate the outstanding generalization performance. The proposed SVM-RBF and SVM-Linear variants have been compared with the traditional algorithms (Maximum Likelihood Classifier (MLC) and Minimum Distance Classifier (MDC)), which are highly compatible with remote sensing images. Furthermore, the MLC and MDC are mathematically modeled and characterized with new features. Also, we compared the proposed improved SVM-RBF and SVM-Linear with the current state-of-the-art algorithms. Based on the results, it is confirmed that proposed variants have higher overall accuracy, reliability, and fault-tolerance than traditional as well as latest state-of-the-art algorithms.

Highlights

As a vital tool for information retrieval regarding land cover and land use, remote sensing (RS) technologies have widely been used in various areas [1,2,3]
Experimental Results This section presents the experimental results in terms of the experimental setup and performance results for the support vector machine (SVM)-radial basis function (RBF) and SVM-Linear, non-destructive and content independent (NDCI) [8], hyperspectral remote sensing (HSRS) [19], SCMask R-CNN [17], computational intelligence applications (CIAs) [18], Key Component analysis (KCA) [21], Addressing Overfitting on Pointcloud Classification (AOPC) [22], MLC [33] and MDC [34]
The results confirm that 99.65% and 99.43% change-detection accuracy has been obtained with SVM-RBF and SVM-Linear respectively; whereas the NDCI, CIAs, SCMask R-CNN, KCA, HSRS and AOPC have obtained the change-detection accuracy 95.6%, 97.4%, 95.0%, 95.8%, 95.2% and 94.2% respectively

Summary

Introduction

As a vital tool for information retrieval regarding land cover and land use, remote sensing (RS) technologies have widely been used in various areas (e.g., land management, and urban and rural planning) [1,2,3]. RS is the method that provides information about events by assessing the data. The Knowledge of land-cover/ land-use is vital in a number of arenas based on the observations done for the metropolitan and regional future planning [5]. The classification can be implemented by algorithms that are either supervised or unsupervised: the former uses pre-labeled data and the latter uses data without labeling. In RS classification, supervised classification algorithms are usually preferred due to their accuracy and practicability [9,10,11]. The statistical distribution of data can severely decrease the accuracy when the data do not follow those assumptions

Objectives

Methods

Results

Conclusion