Automatic Building and Height Determination from Unmanned Aerial Vehicle Data

Abstract

Determination of up to date 3D maps of urban areas and estimation of building height has crucial importance for variety of disciplines such as: city regional planning, architecture, construction industry, population directorates and planning units. In order to track information for building construction speed and illegal constructions, updating building inventories, preparing feasible urban plans 3D maps of urban areas is important to get. Unmanned aerial vehicle (UAV) is promising and suitable for 3D objects detection. These systems create high accurate 3D height maps for buildings and, therefore, can be used to estimate accurate building boundaries in urban areas. In this study, SenseFly eBee RTK Unmanned aerial vehicle (UAV) was used to obtain 3D data. The aerial photographs obtained with the UAV were processed in order to create a three-dimensional point cloud. By processing the point cloud data Digital surface model (DSM) and digital elevation model (DEM), were created. In order to determine the buildings’ height, Normalized Digital Surface Model (nDSM) was formed by removing DSM from DEM. In order to determine building boundaries, high resolution aerial photographs obtained from the unmanned aerial vehicle (UAV) were classified using machine learning algorithms and support vector machines. After classification, we obtained 160 buildings. Then we estimated the buildings floor height for the selected ten buildings.