targets it is possible to roto-translate all the model and geo- 
reference each point of the scan. 
Analysing from a metrical point of view the result obtained with 
this two methodologies, with 11 points topologically estimated 
it has been calculated the gap of the three coordinates. Table 5 
reproduces the statistic of the coordinates difference defined 
with the two methods. 
  
Mean [em] St.Dev.[em] 
X y Z X y Z 
Photogrammetry | 0.0 | -0.2 | -0.3 | 1.6 | 1.6 | 1.7 
Laser Scanner -0.4 | 0.1 0.1 1.8 1.2 1.5 
  
  
  
  
  
  
  
  
  
  
  
  
Table 5 — Gaps statistics of the GPC coordinates surveyed with 
the total station 
The comments of this schedule are: firstly the number of the 
points is less than the GCP used (20) because both in the laser 
model than in the photogrammetric one the measure of some 
points is hard (for bad resolution) instead for some one is 
affected by a blunder. 
The average is more restricted and shows the goodness of the 
geo-reference in both of this two methods with the topographic 
one. In particular about the laser scans, this shows how the 
estimation of the coordinates of the artificial targets, used to 
roto-translate the instrumental reference system, has been done 
with excellent accuracy. 
In the end the standard deviation of the difference is, for the 
coordinates and in both of this cases, of the same order of the 
precision of the GCP, at the confirmation of the uniformity of 
the accuracy of these three survey methods. 
4. TEXTURE MAPPING 
The Texture mapping is the possibility to drape the picture upon 
the solid model. This representation was made with the 
commercial software TexCapture by NRC (National Research 
Centre) of Toronto (Canada). For this application we used the 
Nikon Coolpix 4300 (2300 x 1700 pixels) camera with 8 mm 
of focal objective and 1/1.8 shutter. 
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
TexCapture, like many digital photogrammetric software, 
allows to calculate the internal and external orientation 
parameters of the images and the following “ortho-projection” 
upon the triangulated solid model. 
During the survey we used some cunnings; the images were 
acquired with the similar angle of the laser scans, using a high 
resolution camera (it is important to have a CCD greater than 
3.2 MPixel) and we left the same zoom in order to keep the 
same camera calibrations for every texture. Actually we did not 
use this mean because the calibration parameters were estimated 
for each of the six shots (five on the side and one on the top). 
During the calibration and orientation the parameters to define 
are 15 and in particular: 
6 parameters of external orientation of the camera: 3 
coordinates (Xc, Yc, and Zc) of the camera projection 
centre and 3 rotations of the camera (pitch, roll, and yaw); 
= 3 parameters of internal orientation: principal distance and 
the coordinates x and y of the principal point of the 
images; 
= 6 parameters of camera calibration including: 2 affined 
image parameters to correct for scale difference and non- 
perpendicular of the x and y image coordinates, 2 radial 
lens distortion parameters (third and fifth order), and 2 
decentring lens distortion parameters. 
The software achieved this purpose writing a co-linearity 
equations system for at least 8 homologues points, recognised 
on the solid model and on the images. 
For the Texture mapping of the Baptistery of Cremona, we 
individualised at least 20 points of each scan, chosen in a 
uniform way, not lying on the same plane. At the beginning we 
estimated the 15 parameters in a combined way; subsequently, 
after analysing the meaning of calibration parameters 
estimation, we locked the orientation parameters and set on 
zero the insignificance one. Anyway the principal distance and 
the positioning of the principal point was kept at the value 
estimated step by step. 
After calibrating the camera the TexCapture software allowed 
us to “drape” the images on the triangulated model, deleting 
both the distortions due to the intrinsic characteristics of the 
camera we used (objective distortion, etc.) and the prospective 
distortions. 
  
Figure 6 — Laser scans aligned a) and b) image with intensity of the laser scanner data c) solid model 
  
492 
    
  
     
   
   
   
   
   
     
   
      
     
    
   
   
   
   
   
   
    
   
  
  
    
   
    
    
    
    
   
   
   
  
   
  
  
  
  
  
  
  
  
     
  
   
   
   
Interne 
ie ETC 
As yo 
solid ı 
In blu 
data ( 
zones 
(Ther 
autho 
surve: 
modu 
techni 
Bapti: 
three 
the to 
Certa 
thank 
centir 
proce 
colou 
COITe: 
result 
and tl