id the 
vation 
scale. 
ne to 
ut 60 
netric 
ligital 
or the 
te the 
of the 
yorted 
arable 
ultiple 
g the 
into a 
s and 
yas set 
allow 
red in 
- final 
ration 
"sition 
ent. 
> final 
s that, 
plates, 
of the 
yuided 
1s for 
oor to 
metric 
WS to 
g the 
  
Figure 2 
This camera realises images with a 2552x1920 pixel resolution, 
with a fixed focal lens comparable to a 28mm. Like all the 
photogrammetric cameras, it is furnished with a calibration 
certificate that contains not only the principal distance and the 
position of the principal point, but also the constants A,, A; and 
Ry (Figure 3) used in the formula 
dr = A, (p Rl), (r^ Ri] 
which gives the value of the radial symmetric distortion of the 
lens in order to vary the radius r. 
Caratteristiche della camera: 
Dimensione sensore: 
2552 x 1920 pixel 
8.932 x 6.720 mm 
  
Dimensione pixel: 
0.0035 mm 
   
  
Costanti: 
ck= -7.404 
xe= 0.3060 
yc= 0.0088 4 3 
A1- -0.002277 1: -4.466 ; 3.360 
A2= 0.00005375 2: 4.466 ; 3.360 
R0=3.00 mm 3: 4.466 ; -3.360 
4: -4.466 ; -3.360 
Figure 3 
| 
  
EIL 
  
qj ditaten indt 
  
FAT AG 2 43973 o7 otn 
  
  
  
  
  
r T T 
an e a 39. 45 
dii uml 
Figure 4 
As we can see (Figure 4) from the curve, the distortion values 
cannot be disregarded: at a distance of 1.20 mm from the centre 
of the lens, we find errors of roughly 17 microns, which 
correspond to 5-6 pixel shifts in the image. 
These values compromise the use of this instrument in the sort 
of processing that requires photogrammetric software that 
cannot accept excessively high distortion values. 
The authors implemented the ViaDist software which allows the 
user to upload the image and the parameters of the calibration 
certificate of the camera and consequently correct the errors due 
to distortion (Figure 5). In the event that the values of the 
constants Aj, Ay, Ry are not known, but only the values of the 
distortion calculated in certain points, the programme uses 
polynomial interpolation and calculates the curve of distortion 
of the lens. The new corrected image is created by resampling 
the initial image and moving each pixel of the quantities: 
dun, xs dum. 
r r 
To visualise the errors in every point of the image, a resampling 
function has been inserted that creates an image where the 
distortion values are normalised in the grey scale according to 
their value (black: no distortion, white: maximum distortion). In 
figure 6, we see how the distortion value is zero at the centre of 
the lens and grows up to a certain value of the radius (1.6mm). 
It then decreases again before disappearing in the R, value 
(3.0mm) to grow again up to the edges of the image, in 
accordance with the progress of the curve seen in fig. 4. 
   
  
y «küskTusk?t3"2 «123 ekd*u^4 ekSrx^8 
  
  
  
  
   
  
  
  
Y 7 
40 30 as 
cn jen