hul 2004
well as
essary.
d by the
ject and
hms, by
applied
lication,
storation
xamples
Pseudo-
Filter,
Linear
aggelos,
g finite
. 1995).
ate the
s. Since
camera
may be
st mean
s which
of the
¢ value
ing the
F linear
ditional
ing the
arrays
olution
F) and
(Jahn,
line
slightly
ray can
ittitude
€ shall
optical
plane.
tectors
d with
S given
ès
e
62)
ietrical
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
+ if Dera’
a 2 2
) elsewhere
Ha Si
A double point source &(xo), 6(Xota), a is the distance between
both point sources, generates an spatially dependent intensity
distribution in front of the focal plane H(x — xo) + H(X — xo - a).
H(x) is the system PSF without the pixel PSF. The signal in
the sampling point i-A is obtained by integrating the optical
signal over the pixel area
512
HiA)= [ri ei^ x.) He IA x, a) pis
572
Assuming a Gaussian-PSF with a “width” (standard deviation)
GO, the integration can be performed using GauB’s probability-
integral. The result (see Jahn, 2000) is a pixel dependent
intensity distribution
Abel Ze
a7 3; )
+ di + KG Ma faded d xy ta )
Ac lo gii Niele 8 M
The measured values I(14) depend for a given pixel distance
Alo and a pixel size à/0 (related to the PSF width ©) also on
the point position (phase) xo/o and on the distance a/c between
the light spots. That means that the accuracy definition or the
contrast depends on the displacement of the point source
relative to the pixel.
We introduce as a measure for the resolution the a Minimum
Resolvable Distance (MRD). In the context of a Rayleigh-
criterion based resolution concept, the MRD is the minimum
distance of two radiating points to be resolved.
4 T I 7 r T T T I T T r
r 4]
Pa
E reg
r mei |
5 Diffraction Lirnited Optic pco d 4
3 we : T AT -
— oe = Used Optic a
E whe! ,
wr kt ]
I E E = 2
Le ; Ae MX
e F Linear Array ——> 2i 7
= tT ;
= je permet : Z7 € = Ar =
2 vp 7&8-7- Staggered Array =
= = T
E 7 m di J
| C Resolution (Linear Ar ray/Staggered Array) 4
v : ! UE e eco
0 2 T 4 5 a
gcouss
SPSF Dr]
Figure 3. Double point resolution in relation to the optical PSF
Figure 3 shows the resolution in case of double point resolution
(distance between the two spots with respect to the pixel size)
as a function of the PSF parameter 6 (the used optics has a f-
number = 4). For the diffraction limited optics the equivalent 6
is 1um (see Jahn, Reulke, 1998) and for the real ADS40 optics
we have o « 2.5um.
In case of negligible PSF, the limit for the double point
accuracy is valid. With increasing PSF-width c especially for
the staggered arrays, the MRD also increases, while the
resolution for single arrays keeps constant. We find the
minimal MRD for staggered arrays in case of diffraction
limited optics. Increasing o further the MRD becomes worse.
Typical values for resolution are in the range of MRD « 3*6 or
for the ADS40 camera MRD = 1.5 pixel.
The resolution can be improved by image restoration.
Figure 4. Resolution improvement of restored images
The left image of figure 4 shows an example used for
resolution derivation from figure 3. With a 0z0.5 pixel the
minimal resolution in the sense of the Rayleigh criteria is
about 1.7 pixel. After applying the restoration algorithm the
resolution can be improved by about 3096, which is equivalent
to an improvement of the system-PSF.
3. PROCESSING OF THE DATA
The following chapter describes the data reception and
processing.
3.1 Processing scheme of ADS40 data
After archiving images, position data and other house-keeping
data, the GPS/IMU data is processed with the GPS base station
data. This results in position and orientation files which are
used to create Level 1 rectified images, which are stereo
viewable, ready for processing in many classical remote
sensing systems and are used to perform triangulation,
compilation, DTM production, etc. Further image analysis
products and level 2 rectified greyscale, colour and
multispectral orthophotos can also be created.
After the GPS/IMU data has been processed, the position and
attitude files and a simplified interior orientation of the camera
arc used to rectify the images to a ground plane at a user-
specified elevation. This allows the correction of the aircraft
motion and results in stereo viewable images (Level | rectified
images). Each rectified image is broken up into large blocks of
a user-specified size and can be written in standard formats
such as 8-bit TIFF, 16-bit TIFF and tiled TIFF. Furthermore, it
is ready for human stereo viewing and point measurement
using image matching techniques and other automated
processes.
Acrial triangulation is used to combine the short-term accuracy
of the IMU with the high global accuracy of GPS. In
combination with a minimum number of ground control points,
aerial triangulation delivers best fitting results on the ground.
The extra information added to the system by tie point
measurement leads to very reliable orientation results.
3.2 Data base
For data evaluation we concentrate on a flight over the
Rheintal area, which covers a flat region without breaks. This
test site is located close to the Leica facilities in
Heerbrugg/Switzerland and is typically used for the in-flight
calibration of the ADS40 sensor system including camera
interior orientation parameters and the spatial relation to
GPS/IMU components. The typical calibration flight geometry
is similar to the flight pattern of this test, where several
crossing strips were collected in different height levels to
investigate resolution potential of this data (see figure 5) and to
achieve a rigid block geometry for calibration tasks. The
