Object-Based Analysis and Fusion of Optical and SAR Satellite Data for Dwelling Detection in Refugee Camps

Abstract

This study investigates the potential of very high spatial resolution (VHSR) optical WorldView-2 (WV-2) and single-polarized TerraSAR-X (TSX) synthetic aperture radar (SAR) satellite data for an automated detection of different dwelling types in a refugee camp by means of object-based image analysis (OBIA). First, the optical data and SAR data are analyzed independently, and then a fusion of both data sets is performed applying two different approaches: 1) an overlay operation-based procedure integrating the independent results of the optical- and SAR-based dwelling detection, and 2) a feature-based analysis approach taking advantage of the conjoint analysis of both data sets. The results of the single-sensor and the data fusion approaches are evaluated in detail on the basis of object-based and area-based accuracy assessments. Advantages and limitations of the analysis approaches are discussed. The accuracy rates reveal that the use of optical satellite data shows promising results regardless of the dwelling material, while the SAR data are suitable for the detection of metal sheet dwellings only. In complex camp areas, with closely spaced containers, the results of the independent analyses can be improved significantly by the proposed fusion approaches. The combination of SAR and optical data allows for the separation of contiguous dwellings in cases this was not possible by the optical sensor information