International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
Feature Extraction And Modeling Of Urban Building From Vehicle-Borne Laser Scanning Data 
B.L Lit Q0. Li. 7. Shi, F.F.Wu 
(1. Center of Spatial Information and Network Communication, Wuhan University, 430079, Wuhan, PR China, 
lee(@)wtusm.edu.cn 2.Advanced Research Center for Spatial Information Technology, Department of Land surveying 
and Geo-Informatics, The Hong Kong Polytechnic University, Kowloon, Hong Kong, China;) 
KEY WORDS: Laser Scanning; Algorithms; Feature; Modeling; Extraction; Urban 
ABSTRACT: 
Laser scanning is playing an important role in an acquisition of the 3D spatial objects in urban because of real-time and high 
efficiency than traditional methods. The research presented in this paper is focused on segmentation of range image, especially 
captured by vehicle-borne laser scanning systems, which is the crucial technology for extracting objects from laser scanned data. 
Here, a method, named as Density of Projected Points (DoPP) is proposed for the range image segmentation. With such a process, 
it is possible for feature extraction, object classification and modeling without other auxiliary data. An experimental study is 
conducted to demonstrated the feasibility of the proposed solution on urban building object extraction. 
1. INTRODUCTION 
Geo-spatial information technology is one of the focuses of 
research of worldwide countries nowadays, and the acquisition, 
processing and application of information are the three themes. 
The technology of high speed acquisition and automatic 
processing of spatial data is a crux need to be solved urgently in 
the field of Digital Earth. In the late 1980s, laser-scanning 
technology gained important improvement in the real-time 
capture of 3D spatial information, which provided an option for 
the acquisition of high resolution spatial information, and 
enabled people to capture data moving from point by point 
traditionally to continuous automatically. Laser-scanning 
survey technology is not restricted by weather, and it directly 
captures high precision 3D data in uncontacted active survey 
way. It has superiority which tranditional survey technology 
cannot substitute,and it has been developed as an important 
supplement of photogrammetry and remote sensing. 
Based on the difference of mounting platforms, laser-scanning 
systems were classified into Airborne Laser Scanning System 
(ALSS; also called Laser Range Finder, LRF; or Airborne 
Laser Terrain Mapper, ALTM) [Flood 1997, Killian 1996] and 
Ground-based Laser Scanning System (GLSS; or Vehicle-borne 
Laser Mapping System, VLMS) [Li 2000,Li 2003]. 
ALSS is a multi-sensor integrated complex system, and it has 
been gradually matured [Ackermann 1999]. Currently in the 
world there are some systems as follows: the TopEye system 
from the Saab Survey in Sweden, ALTMS from Optech in 
Canada, the Fli-Map developed in the U.S. by John E. Chance, 
and TopSys and  TopScan from Germany. Airborne 
laser-scanning technology has extensive application in surveys 
of sagging power lines [Wehr 1999], and surveys of coastlines 
and forests [Kraus 1998], mine fields, and urban areas [Kraus 
1998, Erik 1997, Haala 1997, Murakami 1998]. 
The products and application of the GLSS for data capture are 
still at developing stage. Based on the style of data capture, it 
can be divided into both static-scanning and mobile-scanning. 
The typical product of static-scanning is the Cyrax serials from 
Leica. Mobile scanning system such as the sample vehicle 
developed by Tokyo University [Manandhar 2001], which 
consists of three laser scanners and six line cameras, is mainly 
applied in the data capture and modeling of city buildings and 
roads. Hi-Tech Research and Development Program of China 
(Project 863) has successively funded — vehicle-borne 
laser-scanning survey technology [Lu 2003]. 
Data captured by laser scanning are called "Range Image" or 
"Points cloud" which consists of discrete vector points. The 
range image captured by close-range and high-resolution 
consists of abundant features: both topographic and ground 
objects information. So it is possible for us to extract feature 
according to height information besides constructing 
Intern 
high« 
Becat 
topolc 
brings 
proble 
consti 
scann 
laser | 
comp 
resear 
But al 
buildi 
methc 
image 
CCD 
are m 
constr 
for in 
1999] 
cost ( 
compl 
Segme 
model 
technq 
range 
2.1A 
Veh 
combi 
attitud 
Odom 
the ve 
at the 
of the 
gives 
attituc 
the T 
deterr 
object 
the 31 
a