Building Damage Detection Based on OPCE Matching Algorithm Using a Single Post-Event PolSAR Data

Synthetic aperture radar (SAR) is an effective tool in detecting building damage. At present, more and studies detect damage using a single post-event fully polarimetric SAR (PolSAR) image, because it permits faster convenient detection work. However, the existence of non-buildings obliquely-oriented buildings disaster areas presents challenge for obtaining accurate results only PolSAR data. To solve these problems, new method proposed this work to completely collapsed full polarization image. The makes two improvements detection. First, provides solution non-building area removal images. By selecting combining three competitive features, can remove most effectively, including mountain vegetation farmland areas, which are easily confused with buildings. Second, significantly improves classification performance standing A feature was created specifically via development optimization contrast enhancement (OPCE) matching algorithm. Using developed combined texture effectively distinguished buildings, while simultaneously also identifying affected error-prone areas. Experiments were implemented on datasets obtained mode: Radarsat-2 data from 2010 Yushu earthquake China (resolution: 12 m, scale study area: 50 km2); ALOS PALSAR 2011 Tohoku tsunami Japan 23.14 113 ALOS-2 2016 Kumamoto 5.1 5 km2). Through experiments, proven obtain than 90% accuracy built-up extraction achieved accuracies 82.3%, 97.4%, 78.5% Yushu, Ishinomaki, Mashiki town sites, respectively.