Object oriented assessment of damage due to natural disaster using very high resolution images

A building damage assessment method applied to the case of the earthquake of Bam is proposed in this paper. It uses two very high resolution images and focuses on the footprints of the buildings. The need of an accurate registration of the buildings is demonstrated; a registration method that improved the damage assessment is proposed. It allows a classification performance of the buildings among four damage grades up to 78%. The impact of a lower accuracy of the buildings roofs segmentation is evaluated; we show that it mainly leads to a decrease of the capacity to identify the partial damage on buildings.