tituto Tecnico 
NRC Canada, 
by means of 
acy, a photo- 
f thie project. 
elative simple 
surveying and 
io perform the 
or this project 
) team created 
RST team and 
its texturing. 
rveys will be 
and rendering 
results of a 
are presented 
laser scanner- 
ruments werc 
Mpixels digital 
Is), has been 
ded a limited 
fish-eye lens), 
valls, 8 for the 
. The average 
wall per pixel. 
purposes it is 
ser of required 
conditions and 
"hen, the Riegl 
loyed to create 
im is provided 
tating mirrors, 
of view (FOV 
eed of several 
nging between 
listance) with a 
sides X,Y, and 
r the reflected 
Coordinates of ground control points, as common points for the 
comparison between the two models, were measured with a 
Leica TCR 705 total station, featuring a EDM with two laser 
beams: one, invisible, for conventional measurements to prism 
and a second one, visible, for reflectorless surveys. About 70 
control points were measured with the total station with an 
accuracy of about 5 mm. In order to compare the geo- 
referencing of the laser scanner model using both natural and 
artificial targets, part of the control points (50) were chosen as 
easily recognizable features on the frescoes, whereas further 20 
retro-reflective targets were placed mainly at the bottom of the 
walls of the room to avoid any damage of the frescoes, and 
were surveyed in reflectorless mode (see figure 1). For an 
exhaustive overview of both the Leica total station and the 
Riegl laser scanner specifications see Tables 1 and 2. 
Table 1. Leica Total Station specifications 
  
  
  
  
MANUFACTURER Leica 
PRODUCT TCR 705 
Angle measurement 5^1 5 mgon 
Distance 3000m (with reflector); 2mm * 2ppm 
measurement 170m (w/o reflector); 3mm* 2ppm 
  
«1s (with reflector) 
Measuring time typical 3-6s (w/o reflector) 
  
  
  
Recording 78000 measurements and coordinates 
232 interface for external connection 
Magnification 30 x 
Plummet Laser: located in alidade, turning with the 
  
  
instrument. accuracy + 0.8mm at 1.5m 
  
  
Table 2. Riegl laser scanner specifications 
  
  
  
  
  
MANUFACTURER Riegl USA 
PRODUCT LMS 7360 
Laser Wavelength (in nm) 904 
Laser Power (in W, mW) 1 mw 
FDA Laser Classification (Class) ! 
  
Beam Diameter at Specified Distance 
(0.Y ft at X ft/Ymm at X m) 
20 mm at 50 m 
  
  
  
Measurement Technique LiDAR 
Average Data Acquisition Rate (pps) 8,000 
Maximum Data Acquisition Rate (pps) 12,000 
  
Distance Accuracy at Specified Distance 
(0.Y ft at X ft/Ymm at X m) 
Position Accuracy at Specified Distance 
(0.Y ft at X ft/Ymm at X m) 
& mm at 200 m 
  
& mm at 100 m 
  
  
  
  
  
Angular Accuracy (degrees-min-sec) 0.002 
Minimum Range (feet/m) im 
Maximum Range (feet/m) 300m 
Field of View (vertical angle) (degrees-min-sec) 30 
Field of View (horizontal angle) (degrees-min-sec) 360 
  
Minimum Vertical Scan Increment 
(degrees-min-sec) 
Minimum Horizontal Scan Increment 
(degrees-min-sec) 
0.002 
  
0.002 
  
  
  
  
3. IMAGE-BASED MODELING 
The modeling approach adopted at this stage is based on a 
semi-automatic technique described in [El-Hakim et al, 2003] 
and implemented in ShapeCapture commercial software 
[ShapeQuest, 2004]. The bundle adjustment in the software 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
was done with free network, i.e. no control points were used at 
this stage and the model scales were determined from a small 
number of linear measurements (such as window dimensions) 
collected while taking the images. The resulting geometric 
models are shown in figure 2. 
  
  
   
      
   
   
   
       
    
    
—Á 
N 
Room Interior 
Entrance 
Figure 2: Wire-frame models by image-based method. 
4. TEXTURE MAPPING 
Achieving high level of photorealism was another goal of the 
project, in order to create extremely realistic visual 
experiences. Due to the nature of this frescoed room, in this 
stage of the project, cffort has been spent to identify the main 
factors affecting the visual quality of the models and the 
performance of interactive visualization and to select the most 
appropriate technology. Basically, the modelling and rendering 
approaches used for this project can be summarized as follows: 
l. Create geometrically accurate model with an image-based 
photogrammetric technique. 
Divide the model into efficient size groups of triangles. 
Select best image for each group. 
Compute texture coordinates of vertices using internal and 
external camera parameters. 
IS