ER 
dzka 53, 
models are 
d terrestrial 
ler to model 
?. TLS and 
SLR camera 
oftware and 
inal models 
models and 
uilding was 
ments were 
e as follow: 
on airborne 
arise at the 
the different 
tal accuracy 
odel on the 
performed. 
bined point 
ydelling has 
et al., 2010) 
odel on the 
Vosselman 
the basis of 
> horizontal 
, this is also 
he roofs of 
man, 2011) 
ffecting the 
essment of 
(mainly the 
acy is based 
sulting from 
data. 
sented. The 
| of selected 
ent of these 
based on an 
lata and is 
man, 2011). 
of accurate 
ference data 
To proceed 
iption of the 
   
  
  
  
    
  
   
  
  
  
  
  
   
   
  
  
  
   
  
   
   
    
   
    
    
  
   
   
   
   
  
   
   
   
   
   
   
   
  
   
  
  
  
  
  
  
  
  
  
   
   
  
   
  
   
   
   
   
   
  
   
   
  
   
data used for modelling and the various stages of modelling 
have to be presented. 
2. INVESTIGATION OBJECTS AND DATA 
2.1 Objects 
3D modelling of buildings was performed through a project 
commissioned by the regional authorities in south-west Poland. 
The aim of project was to develop three-dimensional models of 
historic buildings in several cities of south-west Poland. The 3D 
models have been created for regional promotion on the Internet 
and also for internal local authorities relating to the protection 
of heritage buildings. Thus expectations defined by the users of 
models require a compromise between modelling detail and the 
ability of models presentation on the Internet. The accuracy and 
level of details modelling were established between LoD3 
(Level of Detail) and LoD4 (Open Geospatial Consortium, 
2012). The idea was to model as faithful as possible 
reconstruction of the building elements and architectural details, 
while maintaining the possibility of effective presentation of the 
model on the Internet. For the modelling the combined with 
a terrestrial and airborne laser scanning point cloud was used. 
In the presented study the accuracy assessment of modelling 
was performed on two objects, referred as "Object 1" and 
“Object 2”. These objects are located in different Polish cities 
(Fig. 1). 
Data both ALS and TLS on each object were obtained with the 
same sensors, but at different times and elaborated 
independently. Influence of the main source of error, i.e. 
GPS/INS system therefore differs in both places. 
"Object 1" is a castle, enlarged and modernized over several 
centuries. This object is characterized by a mix of architectural 
styles and wealth of detail to be modelled. 
"Object 2" are the buildings of the city centre, around the 
historic marketplace. Historic buildings here also have 
a complex structure and wealth architectural details. 
  
A à : : a : 
Figure 1. Investigated 3D models. Upper: “Object 1" (City 
Brzeg). Lower: “Object 2” (City Opole). 
2.2 TLS data 
Terrestrial laser scanning was performed by two Leica scanners: 
ScanStation 2 and ScanStation C10 with hardware accessories 
and software Leica Cyclone Scan 7.1. For the purpose of 
scanning it was set up a few warp points around the objects. 
These points were measured with use of GNSS technique 
supported by ground based augmentation system of ASG- 
EUPOS. For measurements the receiver Trimble R6 was used. 
The coordinates of points were determined with the precision 
better than three centimetres. All the coordinates were 
determined in the coordinate system EPSG:2180 and the Polish 
system of normal heights. 
Warp points were used for georeferencing and connection of 
single scans. Terrestrial laser scanning was carried out using 
resection method with the scanner orientation to the targets 
which were set up on warp points. The object scanning was 
made from many positions, with an average resolution of 2cm 
on the object (the average distance between points on the 
object). The combination and transformation of scans were 
executed with use of Cyclone Register 7.1 software. Accuracy 
achieved during registration was better than one centimetre, for 
both components: horizontal and vertical. 
2.3 ALS data 
Airborne laser scanning was performed with the use of Lite 
Mapper system, based on the full waveform scanner Riegl 
LMS-Q680i. Scanning was performed with a nominal 
resolution of 6 points per square meter. Because of the adjacent 
scans sidelap up to 5096, the resulting cloud of points for 
a modelled object has a resolution of 12 points per square 
meter. The final coordinates of the points were delivered in the 
same coordinate systems as the coordinates of terrestrial 
scanning data. 
The produced from the two sensors combined point cloud 
which is the basis for modelling was visualized in Figure 2. 
  
Figure 2. Example of the combined point cloud (part) 
of terrestrial (black) and airborne (red) laser scanning data. 
2.4 Digital images 
The assumption for the project was that final models had to be 
textured with realistic images. In order to fulfil this assumption 
a set of digital terrestrial images was taken with use of SLR 
Canon 40D camera. To avoid the need for textures mosaicking, 
it sought to every plane of model was presented on a single 
image only. For texturing the invisible from the ground surfaces 
of roofs or roofs with a small slope, the aerial images were used. 
These images were taken during the airborne scanning and its 
GSD was about 0.10m.