The camera positions can be shown in this map as the corners 
of a square — the size of the square defines the distance between 
the cameras, and also the intersecting angles of the cameras 
optical axes. Defining the optimal size and position of this 
rectangle requires further calculations, but this visibility map 
greatly helped the design of the camera stand and mount. 
  
Figure 5. Adjustable camera mount 
3. DEFINING PARAMETERS OF THE MOBILE 
TEST FIELD BASE NETWORK 
The concept of the basic configuration of close range 
photogrammetry networks was introduced by Mason. The 
concept essentially means that photogrammetrists develop 
readily usable photography and processing methods for various 
task types (networks) occurring in practice and also specify the 
result characteristics to be expected. We extended the set of 
basic network types with the type of the mobile test-field of 
photogrammetric measurement for face reconstruction and 
specified the function describing the design factor q, 
characteristic of this network. We specified this value in 
function of the average error of scale and image coordinates. 
A measuring field with calibrated control points and the 
calibration of the cameras are necessitated by the task. A close- 
range photogrammetric test field was built; it consists of 
cylinders of various heights mounted on an alloy plate 
(^manhattan-type"), where the points specified are represented 
by the holes located on the cover circles of the cylinders. The 
geometric features of the test field were determined by a Zeiss 
Opton 3D coordinate measuring machine. This instrument 
specifies the coordinates of vertices with sharpness within the 
0.1 micron range and median error of less than one micron. 
The overall precision of a photogrammetric capturing device 
depends on the networking aspects. As we chose a convergent, 
multi-station photogrammetric network, an initial indicator of 
the precision of triangulation in is given by the formulae 
(Fraser, 1984): 
9, - So=—Ldo, (D 
k 
Se 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B3, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
Where 6 = RMS value of X Y Z object point coordinate 
standard error, 
S = scale number, S=d/c, 
q = design factor expressing the strength of the 
basic camera station configuration 
k = number of exposures at each camera station, 
o,= angular measurement resolution 
o = image coordinate standard error 
Given the assumption of k=1 (no redundancy by multiple 
exposures) the equation (1) can be rewritten as: 
q = 0c/(0S) (2) 
Equation (2) shows that knowing the image coordinate standard 
error, the q factor of a network can be given by the object space 
coordinate RMS deducted from experimental error. 
To get experimental errors, a smaller manhattan-type test field 
was placed into the actual one. (figure 6) 
  
Figure 6. Test fields 
This smaller test field was measured by the same coordinate 
measuring instrument, hence the coordinates are known at one 
micron precision. The known points of this smaller test field 
were measured by the photogrammetric capturing device, and 
the difference between the known and measured points gave us 
the experimental error. The measurement has been carried out 
with a Direct Linear Transformation-based Software developed 
by our Department (Molnar, 2010). The RMS error of this 
photogrammetric network from 20 measured points was 0.48 
mm, so the q factor of this mobile test-field with the described 
arrangement can be given as: 
q - 480/ (oS) (3) 
where the precision of the image coordinates are given in pm. 
4. CONCLUSION 
At this stage of our project, the aim of our work was to develop 
and build a capturing device for photogrammetric face 
measurements considering the photogrammetric network design 
factors. Investigation of the network design constraints resulted 
that we chose a four-station convergent photogrammetric 
network. Visibility modelling was performed to be able to 
define the main parameters of the camera stand and the resulted 
    
   
    
   
   
    
     
  
   
   
    
    
    
   
  
   
    
   
  
  
  
  
  
  
  
  
  
  
    
   
   
    
   
   
   
   
   
   
   
    
  
    
  
    
   
  
  
   
    
  
  
visibilit 
further. 
network 
expecte 
photogr 
the scal 
Blanz, 
Synthes 
annual 
techniqi 
York, p 
Fekete, 
fotograr 
12-23. 
Fraser, 
topogra 
Remote 
Fraser, ( 
Close F 
Publish: