While images have been acquired from planes, satellites or the ground since decades, only recently the acquisition from Unmanned Aerial Vehicles (UAVs) has become an additional option. This paper focuses on the automatic processing of images from small UAVs together with terrestrial images from orientation to the functional modeling of buildings. It is shown how highly precise orientation can be achieved fully automatically only based on images also for large unordered sets of wide baseline images. The orientation is used for high quality fully 3D reconstruction from possibly
many high resolution images employing efficient statistical modeling in voxel space based on octrees and a Total Variation (TV) based feature for the quality of the disparities. From dense 3D reconstruction functional models are derived in the form of façades, roofs, windows and doors using statistical sampling. Results for several datasets give an impression of the potential of the developed processing chain.
«While images have been acquired from planes, satellites or the ground since decades, only recently the acquisition from Unmanned Aerial Vehicles (UAVs) has become an additional option. This paper focuses on the automatic processing of images from small UAVs together with terrestrial images from orientation to the functional modeling of buildings. It is shown how highly precise orientation can be achieved fully automatically only based on images also for large unordered sets of wide baseline imag...
»