The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
91
Figure 2. The reference block and control points distribution
The availability of good control points is a crucial requirement
for in-situ calibration. The control points were derived from the
LiDAR point cloud using LiDAR intensity as a tool for point
identification, since no conventional control points were
available in this project. An example for derived control point
in the CIR image and in LiDAR intensity image can be seen in
Figure 3. The expected root means square error of LiDAR
points is abouti 10 cm all component (Csanyi and Toth, 2007).
(a) (b)
Figure 3. The derived control points from the LiDAR point
cloud in CIR image (a) and intensity image (b)
The block was flown in approximately 650 m above ground,
corresponding to an image scale 25300 and the height to base
relation is about 10 for an average endlap of 68%. In other
words, the 17.5° view angle in the flight direction with is very
small comparing to the normal analog camera, which is 73.9° at
the 153 mm focal length. The small view angle in the flight
direction requires more images in the flight lines, but more
importantly, it significantly reduces the Z accuracy of the
determined object points.
The different sets of the system calibration parameters were
computed to analyze the estimated system calibration
parameters using the BLUH bundle block adjustment software
from the University of Hannover. The various combinations of
the adjustment runs with different parameters and results are
given in Table 1. The focal length f = 25.966 from the
USGS/EROS calibration report was used as initial value in
Fable 1 for first two approaches. The twelve additional
parameters were introduced to block adjustment in the second
approach (Jacobsen, 2006). The radial symmetric lens
distortions and systematic image errors are determined with
additional parameters introduced into the block adjustment. The
radial systematic lens distortions from both the USGS/EROS
calibration report and the bundle block adjustment are shown in
Figure 4 and Figure 5. The systematic image errors of the
Redlake MS 4100 digital camera were also determined with
introduced additional parameters as shown in Figure 6.
Approach
[pm]
RMS at
Control Points
[m]
X
Y
z
1
Bundle block adjustment
5.5
0.3
3
0.3
3
1.2
8
2
Bundle block adjustment
with 12 add parameters
4.3
0.3
1
0.2
3
0.2
5
3
GPS supported bundle
block adjustment with 13
additional parameters
5.2
0.3
3
0.2
3
0.5
1
4
Bundle block adjustment
with improved image
coordinate and focal length
4.3
0.3
1
0.2
3
0.2
4
5
GPS supported bundle
block adjustment with improved
image coordinate and focal length
5.4
0.3
4
0.2
3
0.5
5
Table 1. Reference bundle block adjustment results in UTM
The correction for the focal length was 123 pm with introduced
additional parameter to the bundle block adjustment in third
approach. The affine model deformation of UTM system in this
test area was causing a 10 pm correction for the focal length
(Yastikli et al., 2005). The remaining part of the correction for
the focal length could be explained by the effect of actual flight
condition such as air pressure and temperature. The additional
parameters for the location of principle point were not
introduced to the adjustment because the strips, which were
flown in twice in opposite directions, were not available in the
reference block. The corrected focal length, however, resulted
in improved image coordinates, including new radial symmetric
lens distortions and symmetric image errors. The traditional
bundle block adjustment and GPS supported bundle block
adjustments were repeated with the improved image coordinates
and corrected focal length in approaches 4 and 5.
R [mm]
RedLake MS4100-RGB USGS
ReaLake MS4100 CIR IN-SITU
Figure 4. The radial systematic lens distortion from calibration
report and reference block adjustment