Full text: Proceedings, XXth congress (Part 5)

    
  
    
   
  
  
  
  
  
  
  
    
   
    
    
  
   
  
  
  
  
  
  
  
  
  
  
    
     
    
   
   
  
   
    
     
     
    
    
  
    
    
      
    
    
     
    
    
   
  
stanbul 2004 
mm 
/ to the edge 
urvey allows 
temporary 2 
  
  
etric faken; 
ight sides of 
cture of the 
gent images 
ice, only the 
solid model, 
m from the 
top side. 
ht (Torrazzo 
t possible to 
its. and of a 
, it has been 
op of which 
307 that for 
; prism; this 
rting points, 
: of the wall. 
f laser shots 
sed to reach 
flectivity) iS 
tivity). The 
he dark side, 
    
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
makes the geometric reconstruction of the target centre easy 
from the software which interpolates the light surface among 
the points falling back on it (some hundreds) with accuracy of 
some millimetres. At last the cross appearing at the target centre 
allows to survey the centre by total station (figure 3). 
  
Figure 3 - Target used for the laser taken and radiometric 
answer of the points. 
The coordinates of the Ground Control Points (GCP) have been 
determined with uncertainty of 1.2 cm in the x coordinates, 1.6 
cm in y, 1.1 em in z, according to the presumed accuracy of the 
photogrammetric restitution and of the laser survey. 
The surveys with laser scanner have been carried out with a 
terrestrial laser Optech ILRIS -3D of the Codevintec srl of 
Milan. This scanner can work up to 800-1000m in the best 
conditions of the surface reflectivity. The field edge is 40? x 
40° while the frequency of the points measure is maximum 
2000 points/s. At a distance of 100 m the laser trace on the 
object is about 30 mm (perpendicular shot) and the accuracy on 
the flat surface is about 3-5 mm on the points. 
For the Baptistery survey 14 scanning have been carried out: 13 
of which at the square level have been done contemporary to the 
photogrammetric survey. The last one, taken from the top of the 
loggia of Torrazzo at about 100m on the square level, has been 
taken after some months. 
3. CREATION OF THE SOLID MODEL 
In photogrammetry the creation of the solid model is made after 
the external orientation of the images and from the consequent 
fixing of some characteristic control points. Connecting these 
points, we created the primitive meshes of the model. By now, 
the photogrammetric algorithms allow, knowing the external 
orientation of the images, to find out automatically both of the 
characteristic points on the object surveyed (thought the 
operators that pick the characteristic points basing on the 
chromatic variation of the image) and the homological points on 
the other images. For this purpose, in the close range 
photogrammetry it's not allow to use the automatically airborne 
triangulation programs because the geometry of the image is 
much less standard than the airborne case. Anyway it’s already 
exists some algorithms based on the epipolar geometry or on the 
field of the parallaxes variation that, after calculating with good 
approximation the orientation of the images (Sansó, 1973 or 
Zeng at al, 1992) helps to calculate the coordinates of the 
homologue points and, consequently to determinate the object 
points and to improve the external orientation of each image. 
In our case it has been used the software Photomodeler 
(Www.photomodeler.com), a program with huge spread that 
allow to calibrate the cameras, to calculate the external 
orientation of the images. to define the coordinates the object 
points, to reconstruct the surfaces and the images overlapped 
(even if, in this case with some limitations). 
It has been defined 507 points observed at least on 2 frames, 17 
of that chosen between the GCP with known coordinates; these 
allowed to define together the external orientation of the 
Mages and the remaining points in a local Cartesian reference 
System tied at the total station points. Moreover it has been 
491 
added other 1274 points close to the particular feature of the 
structure: corners, edges, lines, windows and all the points 
necessary for the construction of the solid model. 
In the end, with AutoCad, it has been reconstructed the surfaces 
and assigned at each one an arbitrary texture (figure 4). It has to 
point out that the roof of the baptistery has been “invented” 
because, as already mentioned, there are no photogrammetric 
images about that. 
Regarding the processing and the modelling of the laser scanner 
data, every scan has got an own reference system relative of the 
survey point, therefore the creation of the solid model is 
dependent by the merging of all the scans made by different 
positions using any one reference system. The software used for 
this application and for processing the data is the module 
IMAlign by PolyWorks (http://www.innovmetric.com); this 
software allow to chose some natural point for linking up 
different scans (such as some recognizable feature on the object 
surveyed, house corners, identifiable geometry, different 
reflectivity of the walls, etc). 
  
   
ml 
Figure 4 — Solid model of the Baptistery generated by the 
photogrammetry images, upon that there has been impose some 
arbitrary texture 
All of this procedure permit to be easier the survey on the field 
and at the same time to be faster for the absence of the artificial 
targets, used only for the georeference of all the model. The 
align of different scans in made individualizing three or more 
connection points present on both of the two adjacent scans. 
These points, who the position is known in their respective 
reference systems, at the beginning, consent to move closer the 
two scans; the software, that used ICP algorithm, realises 
subsequently an automatic research of the “homologue” points, 
performing a spatial rotation and translation without scale 
variation respect to the reference system of the adjacent scan 
and it estimates the 6 independent parameters: the 3 rotation 
around the axis XYZ and the 3 spatial translation. 
In this way with the alignment it has been given the parameters 
estimation of the orientation and generated a roto-translation 
matrix for each scan. This process has been applied on each 
scan and permits the creation of the complete solid model of the 
object surveyed. 
At the end of this process we has been realised a triangulation 
(TIN) to the single mesh and it has been made a georeference of 
all the model to set it in the reference local system tied to the 
points surveyed by the total station. This process has been done 
using some artificial targets, above described, positioned on the 
surface surveyed. 
Once known the spatial position of at least one of this three
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.