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-