Study on the Extraction of Building Damage Caused by Earthquake from Polarimetric SAR Image Based on Improved Freemam Decomposition

Abstract

Synthetic aperture radar (SAR) remote sensing, which has the characteristics of all day, all-weather imaging, is an important data source to obtain earthquake disaster information after destructive earthquake. In this paper, the Freeman polarization decomposition method is improved according to the earthquake induced building damage characteristics featured in SAR image. Then the Wishart supervised polarization classification based on polarization scattering matrix is applied to extract building seismic damage information, and the seismic damage degree evaluation method is established based on street block scale. Based on the full polarization RadarSat-2 image in extreme disaster area of Yushu Ms7.1 earthquake occurred on April 14, 2010, the seismic damage of buildings in Yushu city, Qinghai Province is extracted, and the seismic damage degree in street block scale is assessed. Test is made according to the visual interpretation of building damage from very high resolution airborne optical images and field investigation, the overall accuracy of the seismic damage assessment results is 0.84, and the kappa value is 0.68, which validates that the improved Freeman method is effective and feasible to identify earthquake caused building damage.