1054
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
The differences between the two solutions are well within the
a'ccuracy specifications of the AEROcontrol system.
Figure 6 shows the differences between solution A and solution
C (base station vs. PPP). For the application of the trajectory for
direct georeferencing, the position differences would have to be
considered. The most relevant difference is the height
difference of up to 0.2m.
297.000 298.000 299.000 300.000 301.000
[sec]
f- roll - pitch - yaw [
Figure 6: Comparison between solution A and solution C (base
station vs. PPP).
The effect of the difference in the GPS solutions to the attitude
determination is relatively small. The effect in the resulting
orientation angles would be insignificant for the most direct
sensor orientation tasks.
For the introduction of the trajectory information as additional
measurement into an extended aerial triangulation (Integrated
Sensor Orientation “ISO”), the effect of the difference between
the different GPS processing methods would not be significant.
Max. [mm] / RMS [min|
ANorth
AEast
AUp
Base vs. VBS
6.3/1.1
7.0/2.2
19.4/4.2
Base vs. PPP
37.1 / 12.1
66.4/38.1
214.1 / 139.7
Table 1: Position differences between the different solutions.
Max.
mdeg] / RMS [mdeg]
ARoll
APitch
A Yaw
Base vs. VBS
<0.1 /<0.1
<0.1 /<0.1
0.1 /<0.1
Base vs. PPP
0.5/0.1
0.4/0.1
1.3/0.6
Table 2: Attitude differences between the different solutions.
5. INVESTIGATION OF THE SYSTEM ACCURACY
The empirical accuracy for the three different flight blocks was
obtained from comparison at independent check points in object
space. Their coordinates were re-determined through bundle
adjustment or direct georeferencing and then compared to their
a priori known reference coordinates.
5.1 Quality of reference points test site Vaihingen/Enz
The reference coordinates have been determined through static
GPS base line surveys in summer 2007. The whole
measurement campaign consisted of 7 measurement days within
a three weeks period of time. Two independent survey groups,
each using a geodetic GPS receiver, did the base line
measurements. The GPS reference station observations were
obtained through SAPOS. A virtual reference station was used
to keep the base line length as short as possible. To control the
overall accuracy of base line observations each GPS group has
done repeated measurements of the same distinct point twice
each measurement day, in the morning and after finishing their
daily measurement. Thus overall 28 repeated measurements of
the same point were delivered from the seven measurement
days necessary. Their statistical variation is given in Table 3
and reflects the absolute accuracy of the analysed base line.
Since the other base lines are of comparable length, the
obtained accuracy of about 1cm (std.dev.) for horizontal and
2cm (std.dev.) for vertical coordinates can be transferred to the
remaining control points. Note that these accuracy numbers also
include the repeatability (i.e. re-identification of object point for
each new measurement), besides the pure accuracy from GPS
survey. For later processing (during the AT runs) the accuracy
(std.dev.) for control points was assumed to be 0.02m for all
three coordinates.
AEast
ANorth
AUp
[m]
[m]
fm]
Std.Dev.
0.008
0.009
0.018
Max.Diff.
0.019
0.018
0.037
Min.Diff.
-0.018
-0.018
-0.034
Table 3: Accuracy of ground control points in Vaihingen/Enz
test site from static GPS survey.
5.2 Determination of boresight misalignment
Direct GPS/IMU exterior orientation (EO) measurements have
been done by the AEROcontrol-IId integrated system. The
trajectory solution A (based on the station “0384 Stuttgart”) has
been used for the later bundle adjustments and direct
georeferencing. Before those EO parameters can be used for
direct georeferencing, they have to be related to the two
individual camera heads of the dual head constellation.
The boresight angles for both camera heads have been
determined by comparing the measured GPS/IMU angles with
the angles of a bundle block adjustment done from the available
image blocks. In the bundle block adjustment the images from
both cameras together with the uncalibrated GPS/IMU results
have been used simultaneously.