A Digital City Model (DCM) is an accurate digital representation of city objects in a three dimensional environment. During the past few years, the generation and management of 3D city models has become an important issue and plays a central role in many applications, such as urban planning, tourism, telecommunication, environmental modelling and many more. Due to the unavailability of 3D city models, many photogrammetry and computer science researchers have worked intensively on the development of algorithms to extract and generate 3D objects. These algorithms are either not accessible to public GIS specialists/users, or are integrated in commercial 3D software at a very high cost.

The majority of research projects concerning the 3D reconstruction of city models from different data have focused on the reconstruction of buildings. The technical objective of this research is to develop a method that extracts and reconstructs buildings from LiDAR (Light Detection And Ranging) point clouds. Tools for the representation and analysis of 3D data are integrated in most of GIS software capable of handling 3D GIS data. Using surface geometry to detect and reconstruct 3D buildings model would be of great benefit to GIS users.

This paper presents a method to detect and reconstruct a 3D Building Model from aerial LiDAR data using ArcGIS software. Due to the fact that buildings are man-made objects, they can be distinguished by their sharp edges and roof planes. First, all objects are detected and separated by the quick generation of a rough DTM, which is followed by the conversion of the objects into features. In the second step, an aspect raster is created and reclassified according to the object orientation angle. This is done in four different classes and guarantees the correct representation of roof planes for each building. The result features are filtered by area. Building’s roofs are then located by merging the aspect planes. A Minimum Boundary Rectangle (MBR) is then created for each building roof. The roof planes are used to create different trend surfaces from LiDAR points, which in turn are used to reconstruct the building roof. Finally, the building 3D model is generated using building roof and detected boundary.