inbul 2004
on of the
filters and
adiometric
equivalent
FIFF as a
more than
| in the lab
al geodetic
rd of the
for direct
vealed the
sub pixel
'rotriangu-
ough the
s of both
icy of the
accuracy
orientation
cal system
ted in the
| point co-
year) the
leta data is
automated
Is. For a
ta JAVA
on aerial
cial issues
ne at low
ight zone,
eat dome
ys with a
| exposure
— 55 m/s)
— 40 m/s)
more. This
ing angle.
side lap of
ing.
be image
lditionally
he aircraft
nages the
1 the pixel
> no larger
ge motion
h ground
the lens is
gricultural
scribed in
more detail. Due to different specifications of the projects the
ground resolution ranged from 15 — 55 em.
3.1 Example Poppendorf
The block "Poppendorf" realized at the 14^ of April 2004
consisted of 139 images flown in 8 strips at an altitude of
approx. 2.400 m, thus resulting in a ground resolution of 55
cm/pixel. The block covered an area of approx. 10 * 11.5 km,
see figure 2.
The images were taken for agricultural purposes, in order to
give a farmer an overview of his current crop and biomass
status (precision farming). For agricultural management
decisions based upon this type of imagery the turn around time
between image acquisition and the delivery of orthorectified and
interpreted data is crucial. Therefore the images were geocoded
within three working days based upon the approach of an
aerotriangulation without ground control points. The kinematic
GPS-data of the geodetic receiver was extracted In
postprocessing. The interpolation of the 1 Hz GPS-data onto the
perspective centre was done by linear interpolation. With the
precise information of the position of the perspective centre and
the approximate angular information of the AHRS and the
heading information of the GPS the necessary tie points were
found automatically with the ERDAS LPS 8.7 software.
P
5 :
9-9... 9-9-9-e-9---—- t
o€9--e-.-9-9-9 10-608 3
B-
sg.
b
e--0-e ee 9-9-9-9-9-9.9. |,—--$-8-9-00
9-9e-9-9-9-9-8.9-0.9.9. 9.9 /9-e-9-9
BA BS ®t 9--8—9-0-9-9-——————--77 E
8-9-9-9.-9-9-9-9-9-9-9. FETS. 92. cabe eh AR
EBB BBB Be BPW BBP.
im a
99 8 8 8 8-9. 9-6-$-9-8-6-e-9-..g.
Figure 2: Photogrammetric block Poppendorf (black dots,
perspective centres of images; triangles, ground control points)
For a “true” estimation of the triangulation results without
ground control 16 ground control points (GCP's) were
introduced first as check points and in the second step as
GCP's, see figure 2 for the spatial distribution. The GCP's were
natural points, collected with a geodetic GPS receiver. Table 3
presents the results of the integration of the GCP's in order to
enhance the triangulation results.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004
Table 2: Residuals of aerotriangulation of the block Poppendorf
- with and without GCP's
Residuals [m] at 16 check points for AT without GCP
X Y Z
Std. 0.94 0.84 0.85
Max. 1.34 E75 2.11
Min. -1.61 -0.97 -1.79
Residuals at 16 control points [m] for AT
Std. 0.26 0.28 0.06
Max. 0.40 0.81 0.07
Min. -0.56 -0.27 -0.15
From table 2 it becomes obvious that without any GCP the
overall accuracy of the block is within 1 m or roughly two pixel.
This positional accuracy is well enough for the anticipated
purpose of precision farming. With the full number of GCP's
the accuracy is within the sub pixel range, which is also a good
indicator of the overall stability of the block.
The subsequent orthorectification was based on the national
DTM 25 with a height accuracy of approx. 2 m. For the
mosaiking of the single images the cutlines were defined
automatically and a feathering algorithm was used for a smooth
radiometric transition between neighbouring images.
3.2 Example Laage
On 6" of September 2003 a flight of a 4 km long part of an
avenue with trees on both sides was conducted with a ground
resolution of approximately 12 cm. The purpose of the flight
was to investigate the possibilities to obtain information of the
trees, the street and the surrounding from nadir looking images
as well as from oblique images. Therefore the central flight line
along the street was designed to gather nadir looking data. For
the oblique images the camera was turned around 90 degrees
and held out the window of the airplane manually. On small
aircrafts such as a Cessna 172, the wheels of the aircraft
maintain outside during the flight. Due to this fact oblique
images out of the window could not be taken at the anticipated
45? angle. Instead the looking angle was approximately 60^ in
omega. In order to get an idea of the left and right side of the
street two strips with oblique images were flown, see figure 3
for the flight pattern. To become oblique stereo images with and
end lap of 60% the automatic trigger control of the flight
management system had to be reset accordingly.
Figure 3: Flight pattern for a combined nadir and oblique aerial
survey of a street