Vol. 11, No. 2, 108-122, 2012

Comparative analysis of land cover mapping techniques on a Mediterranean landscape using FORMOSAT-2
Montasser Jarraya, Ioannis Manakos, Chariton Kalaitzidis, and George Vozikis

Abstract
Landscape fragmentation is quite dominant in Mediterranean regions and poses significant problems in semi-automatic satellite image classification methods. The issue is somewhat alleviated when high spatial resolution data are used, allowing the production of detailed classification schemes, using either pixel- or object-based classification methods.

The main objective of this research is the comparison of classification methods for Land Use/Land Cover (LU/LC) mapping using high spatial resolution data provided by the FORMOSAT-2 satellite. Three pixel-based and an object-based classification approaches are evaluated; the pixel-based methods employing the Support Vector Machine (SVM), Maximum Likelihood (ML), and Artificial Neural Network (ANN) algorithms, and the object-based classification using the Nearest Neighbour classifier.

All three methods were assessed and compared to each other with respect to the overall and individual accuracy of their classification results, in order to determine the most efficient method. The comparison was made both in terms of overall classification accuracy as well as in terms of individual class identification accuracy. The differences in the performance of each classification method are discussed.

View Full Text (pdf file, 11.2 MB) previous page
History
Submitted: 12 Dec 2011
Revised: 29 Oct 2012
Accepted: 29 Oct 2012
Published: 20 Nov 2012
Responsible editor: Carsten Jürgens

Citation
Jarraya M, I Manakos, C Kalaitzidis & G Vozikis, 2012. Comparative analysis of land cover mapping techniques on a Mediterranean landscape using FORMOSAT-2. EARSeL eProceedings, 11(2): 108-122

EARSeL-logo

EARSeL European Association of Remote Sensing Laboratories, Strasbourg, France

   
BIS-logo

BIS-Verlag
BIS Library and Information System, Carl von Ossietzky University of Oldenburg

 

ISSN 1729-3782