3D building reconstruction in a remote sensing workflow

Abstract

3D reconstruction and recognition of buildings and urban scenes is crucial for urban planning, disaster evaluation, historical, architectural and/or urban research and other photogrammetric tasks. The purpose of this research work is to evaluate 3D reconstruction of urban scenes by using the simplest workflow given by the optional IMAGINE UAV module for the ERDAS IMAGINE 2016 software. Digital elevation models and point clouds were extracted from RGB aerial images, which were captured by two SVS-Vistek nadir cameras with a resolution of 16 megapixel. The cameras were synchronized with a GPS/INS module for a direct georeferencing of the images. Additionally, the generated 3D models were imported to the Agiosoft PhotoScan application for a better visualization and to verify the focal length of the camera. The tridicon BuildingFinder software was used to improve the quality of the produced point clouds and for building recognition as well.