Construcción de geometrías 3D utilizando imágenes UAV como alternativa de solución para la oclusión fotogramétrica

Abstract

Photogrammetric occlusion occurs in urban areas, making difficult to reconstruct a 3D geometry of an object, as certain areas are not properly identified in an aerial photograph. Correctly studying this phenomenon allows to generate true orthoimages, avoiding the double mapping phenomenon, also called ghost image, mainly caused by the perspective projection. This research work aims to show the results got from the construction and deployment of a geometrical methodology for detection of photogrammetric occlusions and reconstruction of true orthoimages, by using datasets associated to the digital surface model and conventional orthoimages generated through high-resolution imagery processing related to a photogrammetric model. Behavior of the proposed methodology was initially validated on simulated datasets belonging to a controlled scenario and, subsequently, on a controlled scenario that sought to recreate the existence of three-dimensional geometries. Finally, its behavior was verified in a real scenario, through the use of high-resolution imagery obtained for a flyover carried out over the Headquarter “Aduanilla de Paiba”, one which belongs to the Universidad Distrital Francisco José de Caldas, located in Bogotá D.C., Colombia. In all cases, from visual verification carried out for areas detected as occluded, it was possible to obtain coherent results in the reconstruction of the true orthoimage.