Building height estimation in single SAR image using OSM building footprints

Kurzfassung

Automatic detection and 3-D reconstruction of man-made objects in particular buildings from single very high resolution (VHR) SAR image is of great practical interest especially when it comes to applications having stringent temporal restrictions e.g., emergency responses. In particular, estimation of building height is of key issue in such post emergency scenarios aiding information retrieval especially in urban regions. However, due to inherent complexity of SAR images originating due to side-looking geometry and different microwave scattering contributions (e.g., problems such as layover and shadowing), the problem is highly challenging. Existing techniques on this problem can be categorized into two groups: methods that employ radiometric analysis of backscattering model or perform geometric analysis of layover and shadowing. However, for radiometric analysis, the precise backscattering modeling of complex urban scene is often difficult (due to lack of knowledge pertaining to specific materials and surface roughness), while for geometric analysis, the scene complexity in dense urban areas often gives rise to, occlusion issues i.e., interference between buildings with larger overlapping layover regions, rendering difficulties in accurate height estimation. Use of auxiliary information is therefore needed not only in developing operational algorithms but is actually inevitable in precise estimation of buildings height in larger areas. Motivated by this, the objective of this work is to estimate buildings height using single VHR SAR image, with the help of such openly available 2-D building footprints acquired from OpenStreetMap (OSM). Detailed methodology and experimental results of building height estimation over single TerraSAR-X image will be demonstrated in the full paper.