ERA 
alencia, 
librations, laboratory 
ted in the literature. 
ori unstable intrinsic 
sted, though some of 
1eology, for instance, 
architectonic facades, 
door/outdoor cultural 
igital B/W bandwidth 
arameters K, and K». 
s depending more on 
ST SITE 
the Hitachi KP-F2A 
rome charge-coupled 
| has a focal plane of 
hich correspond to an 
ideo signal. The focal 
vertical pixels. Each 
n (H) by 7.4 um (V). 
was used. Each B/W 
xel. The output image 
a ranges from 400 nm 
at 760 nm (Figure 1). 
the extraction of the 
1 (blue), 080 (bright 
ear-infrared) and 093 
  
   
00 1000 1100 1200 
  
a 
Meu m mt E EE 
response. 
22 Testsite 
Experimental tests were conducted at the test-field of the 
Department of Geodesy, Cartography and Photogrammetric 
Engineering. The test-field has 40 natural targets well- 
distributed on the three axis. The targets, object control points, 
were measured by means of classical surveying. The estimated 
standard errors of the control points were 2 mm. 
3. MATHEMATICAL MODEL 
The extended model equations of the DLT are: 
a, X -bY c eZ di 
x+Ax = 
az X +b,Y +c3Z +1 (1) 
a X +b,Y+c,Z +d, 
yt Ay 2———ÉÓÓB——— 
$ a4X +b,Y +c3Z +1 
where (x,y) are the observable digital co-ordinates, (X,Y,Z) the 
co-ordinates of the object point, (a;,5,c;,d;...c3) the eleven DLT 
parameters of the particular image, and (Ax, Ay) the additional 
parameters considered. After some statistical test, the most 
significant additional parameters for our digital video camera 
were the symmetric radial distortion ones. Thus, the additional 
parameters follow the mathematical model: 
^ 
Ax (X -— X; (Kır +Kyr*) 
Ay=(y - 4M + Kır*) Q) 
The elements of interior and exterior orientation were derived 
independently for each image, as proposed by Dermanis (1994). 
4. CALIBRATION IMAGES AND RESULTS 
The set of images of the test-field was obtained from the same 
position and attitude. A total of 30 images were acquired, taking 
five images in sequence for each filter (081, 080, 061, 092 and 
093) and, finally, without any filter. The nominal principal 
distance was always fixed to 8 mm. Figures 2, 3 and 4 show a 
representation of the set of multispectral images. 
  
Figure 2. Test-field images with #8 mm. Filter 081: left image. 
Filter 080: right image. 
  
Figure 3. Test-field images with /=8 mm. Filter 061: left image. 
Without filter: right image. 
   
Figure 4. Test-field images with /=8 mm. Filter 092: left image. 
Filter 093: right image. 
For the whole set of images the DLT transformation was 
applied and the following orientation parameters were 
computed (Lerma, 2002): six exterior orientation parameters of 
the camera (X;,Y,,Z,,0,9,y); and seven interior orientation 
parameters (xy, f, k,0,K,K») 
Next figures show some of the results derived from the CCD 
video camera. In order to shorten the analysis, only the most 
significant figures are shown. In this way, the variability of the 
principal distance f appears in Fig. 5, the displacement of the 
principal point (x,,y,) in Figs. 6 and 7, and the fluctuation of the 
additional parameters K,,K, in Figs. 8 and 9. The computed 
positions of the projection centre are shown in Figs. 10, 11 and 
12. 
  
  
  
  
  
   
   
   
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 3 
INFRA ROJO 
VERDE OSCURO 
— AZUL MEDIO AZUL.CLARO 
IR. FILTER SIN. FILTRO 
  
  
  
Figure 5. Principal distance f (in pixels). 
  
             
    
  
20 21 22 23 24 25 26 27 28 29 30 
13 14 15 16 17 18 18 
4:2:29::59::4 057657 8.9 10 11 12 
  
——- AZUL MEDIO AZUL CLARO «——-— INFRA ROJO IR FILTER. ——SIN FLTRO ———— VERDE OSCURO 
Figure 6. Principal point x, (in pixels). 
—401-