Change Detection in Remote Sensing Images of Damage Areas with Complex Terrain Using Texture Information and SVM

DOI

10.2991/cas-15.2015.54How to use a DOI?

Keywords

change detection; support vector machines; texture feature; earthquake; damage areas; complex terrain.

Abstract

Traditional methods, when applied to change detection in remote sensing images of damage areas with complex terrain, often result in inaccuracy. And it is difficult to select an appropriate threshold, which however can be solved by support vector machines (SVM) method. Conventionally spectral information is put into SVM as its features; however, the experimental results are not satisfactory. Considering the spatial distribution and structure information of the image, we choose texture information as the new feature. In this paper, a change detection architecture based on the inclusion of texture information and SVM is proposed. Three textures including simple texture, Tamura texture and GLCM texture are adopted in the experiments. By calculating and comparatively analysing four accuracy specifications which are detection rate, missed alarm rate, false alarm rate and Kappa coefficient, we conclude with the experimental results that GLCM texture accurately reflects the diversity of the regional spatial distribution. The inclusion of appropriate texture information and SVM can be successfully adopted to change detection of damage areas with complex terrain.

Copyright

© 2015, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Gao Fei
AU  - Zhang Lu
AU  - Wang Jun
AU  - Mei Jingyuan
PY  - 2015/08
DA  - 2015/08
TI  - Change Detection in Remote Sensing Images of Damage Areas with Complex Terrain Using Texture Information and SVM
BT  - Proceedings of the 2015 AASRI International Conference on Circuits and Systems
PB  - Atlantis Press
SP  - 225
EP  - 229
SN  - 2352-538X
UR  - https://doi.org/10.2991/cas-15.2015.54
DO  - 10.2991/cas-15.2015.54
ID  - Fei2015/08
ER  -