Building Change Detection using Coherent and Incoherent Features from Multitemporal SAR Images

In this paper, we propose a novel method for building change detection using coherent and incoherent features of high resolution multi-temporal synthetic aperture radar (SAR) images. Three coherent and incoherent features are used to define pixel-level building change, while initial result is obtained through multi-threshold. After that, initial result is segmented into different areas based on the same time of change. Then, dynamic time warping (DTW) similarity measure is selected for binary classification in each area to separate it into changed and unchanged classes. Experimental result obtained from 10 TerraSAR-X Stripmap SAR images, shows the effective performance of the proposed approach.