5. Conclusions
This paper presented an approach for the detection of buildings that have collapsed in an earthquake using pre- and post-seismic IKONOS stereo image pairs. Since 3D geometric changes (particularly height change) of collapsed buildings can be detected using HRSI stereo pairs, the accuracy of the 3D ground point determination from HRSI was first improved employing an affine bias compensation model based on RPC bundle adjustment. With the refined 3D ground coordinates, two ways of detecting the collapsed buildings were proposed using pre- and post-seismic HRSI stereo pairs. The first was to detect an individual collapsed building by comparing the height differences of corner points of buildings calculated from the pre- and post-seismic IKONOS stereo pairs, and the second was to detect the region of collapsed buildings by calculating the difference between the pre- and post-seismic DEMs generated again from the pre- and post-seismic stereo images. In addition, the pre-seismic DEM based on a pre-seismic topographical map was generated for detection comparison in the study.