Detection and analysis of collapsed buildings after the kumamoto earthquake by using alos2-palsar2 data

In this work, we analyze the collapsed buildings in Mashikimachi, Japan, after the Kumamoto Earthquake. First, we show the results of distinguishing built-up area from natural target area and qualitatively observing the damage. Second, we show a quantitative analysis on the damage level by employing the recently proposed seven scattering model decomposition. Finally, we discuss the possibility for combining the built-up area detection and damage level analysis by using quaternion neural networks.