the number of sensor elements on the stereo than on the nadir nadir stereo 
sa CCD line (Fig. 12). 
i stereo B sorisor 
ipi nadirg elements 
are | s l re p p' 
— | 
f fe o "1 
dai > 
Figure 12: Effect of camera rotation on the resolution fight, direction 
d 2 i RID RE m (perpendiculoritosthe Figure 13: Effect of camera metric properties on 
ecu p ; scanner images obtained by camera rotation: features in 
A angle covered by an object off-nadir images are elongated perpendicular to the 
f focal length of the lens direction of the rotation. Due to the smaller pixel field 
f' changed focal distance due to camera rotation of view the stereo channel image is darker. 
(f‘=f/cos y) 
hi p distance covered on nadir channel ( p = tan À*f) 9 
x m p/ distance as projected onto the rotated focal plate 5 40 
(p’= tan A*f”) = 20 
3 
of Hence, an object viewed by the stereo channels is * 0 
ike "elongated" in v direction with respect to the nadir channel S -20 
for (Fig.13), i.e., a circle in the nadir channel appears as an = 40 
el, ellipse in the stereo channel, with the long axis oriented = J = 
his parallel to the direction of the CCD sensors (v direction). & -60 
reo je i sino 9n ! E 20. 4005, 800. 11200 1600 2000 
ons We used digital image matching to identify and to quantify pixel number (perpendicular to flight direction) 
this type of distortion in the imagery. The stereo channel 1 
of an ET3 image was matched to the nadir channel (Fig. 14). Fi 14: PI f refi 
With the stereo angle being 18.94?, we estimate that eme Plot zs Ie ence vule numbers versus 
elongation of features in stereo channel images with respect the n eene 9 mue ga sample number to matched 
to nadir images should be about 5% (cf. Fig. 12). This IE AR a tepe 
agrees with what is seen in the disparity data (Fig. 14). image: stereo ‘1 channel). These disparity data 
demonstrate the metric properties of the camera, in 
particular the decreasing field-of-view for off-nadir 
pixels (cf. Figs. 11 and 12). 
4.3. Verification of Geometric Calibration 
The camera tests provide the unique opportunity to verify 
A the geometric precision of the instrument. The calibration 
data indicate that the nadir, the stereo 1, and the stero 2 
mo I T T T TT T + | T T | T TT T T | 1T T | 7117 71.1, | T T 1 | 1 T T 1T T T T h T [ T T a 
E Bi 52 R P2 BL IN GR Pl R | Ss 15 20 
= | N 
= \ 0 
5 10 | \ + 10 o 
Ae \ 2 
i - 
© \ e 
KO \ T 
e. | = 
"d \ > 
zes S in. | is 9. 
= 
2 y | e 
of 8.10 \ — -10 = 
els. o u \ Oo 
els S | -15 o 
ess un 1 il. st 1 1 | )1 1 | 1 il | L | | 1 1 Ja Ji sl si 1 il 1 1 1 Ju koa V = 
try cá | I | | | T | | T 8 
the T 60.05 49.8 39.6 10.2 00 -102 -39.6 -49.9 -60.05 B, 
HRSC along track pixel positions [mm] 
Figure 15: Positions of sensor elements of the different CCD lines on a fictitious focal plane with a focal length of 175 
on mm, 
ger 
353 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996