ISPRS Commission III, Vol.34, Part 3A »Photogrammetric Computer Vision“, Graz, 2002 
KNOWLEDGE-BASED BUILDING DETECTION 
BASED ON LASER SCANNER DATA AND TOPOGRAPHIC MAP INFORMATION 
Alexandra D. Hofmann !, Hans-Gerd Maas | André Streilein? 
! Institute of Photogrammetry and Remote Sensing 
Dresden University of Technology 
Helmholtzstr.10 
D-01062 Dresden, Germany 
Alexandra. Hofmann@mailbox.tu-dresden.de 
? Swiss Federal Office of Topography 
Section Photogrammetry and Remote Sensing 
Seftigenstrasse 264 
CH-3084 Wabern, Switzerland 
KEY WORDS: Segmentation, building detection, airborne laser scanning 
ABSTRACT: 
Laser scanner data is being increasingly used to obtain topographical and object height information for mapping and GIS tasks. 
Valuable information can be derived of the terrain and objects of a region. Different methods have been published to segment laser 
scanner data in order to extract information. This paper aims to discuss the extraction of buildings to transfer 2D building data into 
3D building data of a higher accuracy. For this study, only laser scanner data and scanned topographical maps are to be used. The 
study can be divided into two parts: segmentation and detection. Firstly, the region-based segmentation method is used to delimit 
objects. Then a scanned topographic map, filtered and converted into a vector format, is utilised to locate already recorded houses. 
Buildings not contained in the map could be found by combination of segment attributes. The study shows promising results — more 
than 95% of the buildings were detected. 
KURZFASSUNG: 
Laserscannerdaten werden im zunehmenden Maße zur Informationsgewinnung von topographischen und Objekthóhen für Karten 
und Geoinformationssysteme genutzt. Sie bieten wertvolle Informationen zu Gelände und Objekten. Jedoch ist es immer noch 
schwierig und es benótigt weiterer Daten wie digitale Katasterinformationen oder Grundpláne, um passende Informationen aus den 
Laserscannerdaten zu extrahieren. Dieser Artikel behandelt das Thema Gebáudeextraktion aus Laserscannerdaten mit dem Ziel der 
Datenaufwertung topographischer Karten. Als Hilfsmittel steht nur eine gescannte topographische Pixelkarte zur Verfügung. Die 
Analyse besteht hauptsächlich aus zwei Arbeitsschritten: der Segmentierung und der Detektion. Zuerst wird eine 
nachbarschaftsbasierte Segmentierungssoftware angewendet, um Objekte zu begrenzen. Anschließend wird eine modifizierte 
topographische Pixelkarte als Vektordatei zum Detektieren bereits aufgenommener Gebäude verwendet. Noch nicht registrierte 
Gebäude sollen danach mit einer Kombination verschiedenster Segmentattribute identifiziert werden. Dieses Prinzip der 
Informationsextraktion verspricht gute Erfolge — mehr als 95% der Gebäude der Testgebiete wurden erkannt. 
1 INTRODUCTION Upvaluation of map information, in the sense of transferring 2D 
building data into 3D building data of higher accuracy, is a 
The application of laser scanner data has expanded rapidly current topic of interest, since many applications such as urban 
during the last decade. They are well known as an information modelling are based on this information. Now it is of special 
tool in many environmental applications such as urban interest to find a method to obtain 3D information of objects as 
modelling (Palmer 2001, Haala and Brenner 1997), disaster ^ fast and accurate as possible. This paper aims to demonstrate 
management (Steinle and Voegtle 2001), or flood modelling the steps we have made by using laser scanner data for 
(Cobby et.al. 2001). Other applications, such as mapping of ^ pyaluation of map information in sense of building reposition 
corridors and electrical transmission lines, have been presented and modelling. Auxiliary media used, included simple filter 
by Wehr and Lohr (1999). Laser altimetry data has been also methods and scanned topographical maps as they are available 
discussed with regard to building extraction (Forlani and for the public. 
Nardinocchi 2001, Gülch 2001, Hyyppä and Hyyppä 2000, Lee 
and Schenk 2001, Maas 1999, Stilla et.al. 2001, Vosselman and The main idea of this study is to use the raster pixel map as a 
Dijkman 2001). Existing building information has been used for basis to obtain the majority of the buildings in the laser scanner 
building reconstruction as well (Brunn and Weidner 1997, data set and so to have a direct relation between the found 
Haala 1999, Hug and Wehr 1997, Vosselman and Suveg 2001). ^ puildings and their position on the map. Those buildings not 
Vice versa, it also seems that laser scanner data could be very recorded on the map, can be found by using the unique 
useful regarding the upvaluation of map information. characteristics of buildings. The analysis is composed of two 
A - 169