3D modeling of large urban areas with stereo VHR satellite imagery: lessons learned

摘要

This paper discusses the potentials of very high-resolution (VHR) stereo imagery for automatic generation of digital surface models (DSM) and 3D information extraction on large metropolitan cities. Stereo images acquired by GeoEye-1 on Dakar (Senegal) and Guatemala City (Guatemala) and by WorldView-2 on Panama City (Panama), Constitucion (Chile), Kabul (Afghanistan), Teheran (Iran), Kathmandu (Nepal), and San Salvador (El Salvador) were processed following a rigorous photogrammetric approach. The work focuses on evaluating the quality of the DSMs in relation to the image and terrain characteristics. The size of the datasets, the variety of case studies, and the complexity of the scenarios allow to critically analyze the potentials of VHR stereo imagery for 3D landscape modeling for natural hazards assessment.