International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B3, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
AUTOMATED RECONSTRUCTION OF WALLS FROM AIRBORNE LIDAR DATA FOR 
COMPLETE 3D BUILDING MODELLING 
Yuxiang He*, Chunsun Zhang, Mohammad Awrangjeb, Clive S. Fraser 
Cooperative Research Centre for Spatial Information, Department of Infrastructure Engineering 
University of Melbourne, VIC 3010, Australia 
y.hel6@pgrad.unimelb.edu.au, (chunsunz, mawr, c.fraser)@unimelb.edu.au 
Commission III, WG III/2 
KEY WORDS: LIDAR, Three-dimensional, Principle Component Analysis, Segmentation, Feature extraction, Vertical wall 
ABSTRACT: 
Automated 3D building model generation continues to attract research interests in photogrammetry and computer vision. Airborne 
Light Detection and Ranging (LIDAR) data with increasing point density and accuracy has been recognized as a valuable source for 
automated 3D building reconstruction. While considerable achievements have been made in roof extraction, limited research has 
been carried out in modelling and reconstruction of walls, which constitute important components of a full building model. Low 
point density and irregular point distribution of LIDAR observations on vertical walls render this task complex. This paper develops 
a novel approach for wall reconstruction from airborne LIDAR data. The developed method commences with point cloud 
segmentation using a region growing approach. Seed points for planar segments are selected through principle component analysis, 
and points in the neighbourhood are collected and examined to form planar segments. Afterwards, segment-based classification is 
performed to identify roofs, walls and planar ground surfaces. For walls with sparse LIDAR observations, a search is conducted in 
the neighbourhood of each individual roof segment to collect wall points, and the walls are then reconstructed using geometrical and 
topological constraints. Finally, walls which were not illuminated by the LIDAR sensor are determined via both reconstructed roof 
data and neighbouring walls. This leads to the generation of topologically consistent and geometrically accurate and complete 3D 
building models. Experiments have been conducted in two test sites in the Netherlands and Australia to evaluate the performance of 
the proposed method. Results show that planar segments can be reliably extracted in the two reported test sites, which have different 
point density, and the building walls can be correctly reconstructed if the walls are illuminated by the LIDAR sensor. 
1. INTRODUCTION in Section 4. A discussion of the developed approach is 
presented in Section 5, along with concluding remarks. 
Digital building models are required in many geo-information 
applications. Airborne Light Detection and Ranging (LIDAR) 2. RELATED WORK 
has become a major source of data for automated building 
reconstruction (Vosselman, 1999; Rottensteiner and Briese, Automated building reconstruction from airborne LIDAR data 
2002; Awrangjeb et al., 2010). With its increasing density and has been an active research topic for more than a decade 
accuracy, point cloud data obtained from airborne LIDAR (Vosselman and Dijkman, 2001). Since buildings are usually 
systems offers ever greater potential for extraction topographic composed of generally homogeneous planar or near-planar 
objects, including buildings, in even more detail While surfaces (Hug, 1997; Oude Elberink, 2008), significant efforts 
considerable achievements have been made in building roof ^ have been directed towards the development of algorithms for 
extraction from airborne LIDAR, limited research into the automated point cloud segmentation of planar surfaces. For 
modelling and 3D reconstruction of vertical walls has thus far example, building roofs are generally reconstructed by 
been carried out. However, walls are important components of a exploring the spatial and topological relations between planar 
full building model, and without walls a building model is roof segments. 
incomplete and potentially deficient in required modelling 
detail. Yet, in certain applications such as car and personal Segments can be determined by region growing methods, using 
navigation, building walls are more important than roofs in city edge-based approaches, or via clustering techniques. Region 
models. growing approaches start with a selected seed point, calculate its 
properties, and compare them with adjacent points based on 
The main difficulty for wall reconstruction is the typical low certain connectivity measurement to form the region. 
density and irregular distribution of LIDAR points on vertical Vosselman and Dijkman (2001) explored the use of Hough 
façades. In this paper a method for automated extraction and Transforms for planar surface detection. A random point and its 
reconstruction of vertical walls from airborne LIDAR data is certain neighbours were first selected and transformed into 3D 
presented. The automated identification and location of wall Hough space. The point was then adopted as a seed point in the 
points, along with the development of new methods for reliable case where all the neighbours in Hough space intersected into 
segmentation and classification of point clouds has formed the one point. The other strategy of seed selection is RANSAC 
focus of the reported research. These developments are detailed (Brenner, 2000; Schnabel et al., 2007). A comparison of the two 
in Section 3, together with approaches for wall reconstruction strategies has been reported by Tarsha-Kurdi et al. (2007). 
and modelling. Two test sites have been employed to evaluate Normal vectors from neighbouring points also provide crucial 
the developed algorithms and experimental results are presented information for segmentation. Sampath and Shan (2010) 
113