Building Change Detection Based on 3D Co-Segmentation Using Satellite Stereo Imagery

Building change detection using remote sensing images is significant to urban planning and city monitoring. The height information extracted from very high resolution (VHR) satellite stereo provides valuable for the of 3D changes in buildings. However, most existing algorithms are based on independent segmentation two-temporal feature fusion spectral change. These methods do not consider spatial context simultaneously. In this paper, we propose a novel building algorithm Co-segmentation, which makes full use contained stereoscope data. An energy function containing information, constructed. Image morphological index (MBI), obtained by robust normalized digital surface models (nDSM) difference. Co-segmentation divides into changed foreground unchanged background through graph-cut-based minimization method. object-to-object results overlay analysis, quantitative values calculated according correspondence. superiority proposed that it can obtain buildings planar vertical performance evaluated detail six groups datasets. experimental prove effectiveness algorithm.