2004 
  
eters 
| and 
tions 
10WS 
lally 
d of 
cant 
eter 
rcial 
ten 
e.g 
eter 
tical 
ndle 
high 
ated 
the 
d to 
ents 
the 
their 
ions 
ight 
orne 
rior 
no 
GPS 
rent 
case 
ions 
with 
s of 
nted 
ling 
ials) 
sical 
is 
ised 
eps. 
uite 
are 
SS; 
lese 
1501 
cra 
2D 
ing 
lual 
ere 
and 
"his 
rely 
mer 
03) 
‘ket 
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B6. Istanbul 2004 
  
segment covering the before mentioned classes (1) and (2). The 
second and third group is more or less dedicated for high 
accuracy and large format data acquisition. The Vexcel 
UltracamD and the Dimac Systems DIMAC sensor are other 
systems which are related to class (2). Besides ADS40, other 
actual imaging line scanning systems being used for operational 
airborne photogrammetric purposes are relatively seldom. The 
DLR HRSC family and the Starlabo TLS scanner have to be 
mentioned in this context. Nonetheless, other imaging line 
scanners are used in close connection with laser scanning 
systems to support the automatic classification of laser points. 
One representative of such system integration is the Toposys 
Falcon laser scanner system (Toposys 2004). 
4.1 Applanix/Emerge DSS 
The Applanix/Emerge DSS is chosen as representative of digital 
medium format sensor systems. The optical part is based on a 
MegaVision 4092 x 4077 pix CCD array digital back mounted 
at a Contax 645 medium format film camera housing. This 
housing is stabilized using a proprietary exoskeleton to maintain 
a more or less fixed interior camera geometry. The camera body 
itself is rigidly fixed with an Applanix POS/AV 410 
GPS/inertial system, providing full exterior orientation elements 
for direct georeferencing. The dimension of the used CCD 
matrix is 3.68 x 3.67 cm? (9 x 9 um? individual pixel size) 
which is less compared to the size of medium format analogue 
films (typically between 4.5 x 6 cn? and 6 x 7 env). In 
combination with the two available lens systems of 55mm 
(standard) and 35mm focal length (optional) the resulting field 
of view is 37deg and 56deg. Comparing the field of view to the 
geometry of standard photogrammetric cameras (23 x 23 cm? 
format) these values correspond to a normal-angle (41deg, 
30.5cm focal length) or medium-angle (57deg, 21.0em focal 
length) image geometry, respectively. 
The geometric calibration of the DSS is done by terrestrial and 
airborne calibration. Using a calibration cage imposed from 
different angles the interior orientation parameters of the camera 
are estimated, namely focal length, principle point and lens 
distortion parameters. In addition to the camera related 
parameters, the inherent misalignment between IMU body 
frame system and DSS camera frame is estimated. After 
terrestrial calibration the estimated parameters are verified from 
airborne data. Some more details on the applied calibration 
procedure, the software and the overall performance are 
presented in Mostafa (2004). 
42 Zl-Imaging DMC 
The concepts of the Zl-Imaging DMC system were firstly 
introduced to the photogrammetric users community during the 
Photogrammetric Week 1999. The official market introduction 
took place during the ISPRS congress 2000 in Amsterdam. This 
digital sensor is based on a multi-head solution using four larger 
format CCD frame sensors (7k x 4k pixels, pixel size 12 x 12 
um?) for the slightly tilted pan-chromatic high resolution 
camera heads. From the overlapping images a new image is 
calculated representing an perspective virtual image recorded by 
a large format 13824 x 7680 array. This virtual image is claimed 
to be free of any distortions, hence the knowledge of interior 
orientation of each individual camera head and the relative 
orientations between the different cameras is essential within 
the generation of the virtual image. The applied calibration 
process is divided into two steps: single head calibration and 
y 
platform calibration. The approach is given in detail in Dórstel 
et al (2003), Zeitler et al (2002) and should be recalled here in a 
condensed form. 
4.2.1 Single head calibration 
The lab calibration of the individual camera heads is done with 
the goniometer measurement device available at the Zeiss 
Camera Calibration Centre at Oberkochen/Germany. This 
calibration unit is typically used for the calibration of analogue 
RMK airborne cameras. The goniometer is based on the Zeiss 
theodolite Th2 providing an accuracy of 1 are sec which results 
in an image accuracy of 0.6 um or 1/20 pixels assuming the 
nominal focal length of 12cm for the PAN camera heads. In 
contrary to the classical calibration, which was already 
described in Section 3, the CCD array — rigidly fixed into the 
camera head — cannot be exchanged by a master grid plate. This 
does not allow the measurement of reference points on the 
master grid and the correct auto-collimination of the system. 
Hence, the projected images of the theodolites cross-hair are 
measured in the digital imagery via automatic point mensuration 
approaches. The goniometer measurements are done in four 
different planes (horizontal and vertical bi-section, two 
diagonals), where all measurements in each plane are done 
twice with approx. 180deg rotated camera head. Since this 
rotation is slightly different from the nominal 180deg value and 
the auto-collimination cannot be guaranteed, additional three 
degrees of freedom (3 unknown rotation angles) are introduced 
in the subsequent calibration adjustment, which are estimated as 
unknown parameters for each measurement plane. These angles 
are describing the individual rotation between pixel- or image 
coordinate system of the camera head and the object coordinates 
realized by the goniometer for each measurement plane. 
The desired calibration parameters are determined via bundle 
adjustment, where the calibration terms are estimated as 
additional parameters. In order to use the bundle approach, the 
goniometer angle measurements are transformed into "object 
coordinates" obtained via intersection of the measured rays with 
a virtual plane with constant height. Within the DMC 
calibration the physical relevant parameter set proposed by 
Brown slightly modified as given by Fraser (1997) are 
implemented. Besides the three geometric parameters of interior 
orientation Ax, Ay, and Ac, the first two (Kl, K2) of the three 
radial symmetric parameters are always significant. In some 
cases the affinity and shear terms B1 and B2 are also estimated 
as significant. Due to the high quality lens manufacturing the 
tangential distortion parameters P1 and P2 are non present and 
eliminated typically. The accuracy 6, after parameter 
estimation is about 0.15 pixel or 1.8 pm, respectively. 
Repeating the calibration after certain time interval shows high 
stability of the individual camera heads. The maximum 
corrections after re-calibration are documented with 1/10 of a 
pixel (Dórstel et al 2003). It should be mentioned that the single 
head calibration parameters refer to the "preliminary" single 
head images only. Their knowledge is essential for the 
calculation of the virtual image but they must not been applied 
on the composed images when using these virtual images for 
photogrammetric data evaluation, which should be the standard 
way for DMC image data processing. 
The result of the camera lab calibration is documented in one 
calibration certificate for each camera head. Within this 
protocol, the estimated values of calibration parameters and 
their accuracy (STD) are given. Additionally, the applied 
distortion model formula and some general remarks are 
mentioned. The certificate consists of three pages. 
207