3. Istanbul 2004
3
directions
\
AN
| 3 in A) along-
vation.
N
directions
N
2 'N
à :
N © =
eo
Q9
KM ND
> E
Be x € * N
60
4 in A) along-
vation.
directions
Py
= - e
> Ng X
i A C) = -
5 in A) along-
vation.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
o9. o9. vet.
ras Ton Tes x
\ N x © N N
+ SC ww 4 f \ „Ne
Sa Ne Sec flight directions
x >. N N N N
\ x N X N X 6
NC A NN x N 2
N x S: NN N \
NNN aN NEN N
NN NN
X NS N n N No9 ^
a o^ x 9 x
x ^e x © x
a No N N e? N
un vA Te T?
N X. x N X x
0.1 m WAS eo, Rope
Ra ves vy.
difference x KENNEN
NOM CN AN CN,
© a x e*. 9 ^
Figure 11. Differences between laser strips 2 and 6 in A) along-
track, B) in across-track and C) in elevation.
5. DISCUSSION
The test sites for orientations (Figure 2) were chosen randomly
without ensuring beforehand the suitability of the tie features
for orientations. During the orientation process, it turned out
that in many areas the quality of tie features was inadequate in
across-track direction. In most of cases, the problem was the
orientation of the features. In some cases, the size of the test site
was insufficient, causing the lack of interpretable features.
However, the interactive orientation turned out to be suitable
method to detect even small differences between point clouds, if
the target area included enough visible tie features.
This research focuses on repeatability. Therefore, when the
results are read, it must be remembered the LIDAR can measure
some targets repetitively in an incorrect way. For example, the
material of the target can cause systematic bias (Hyyppà &
Hyyppà, 2003). Nevertheless, it is important to ensure the good
repeatability before any target-based corrections are applied.
The measured differences between laser strips concern the
entity within small test sites. Therefore, a repeatability of single
laser measurement cannot be directly derived from the results.
According visual impressions during the orientations the
repeatability of details vary a lot. The most crucial parameter
seemed to be the gap between scanning strings (Figure 1),
because small details are modelled from different planimetric
location, leading the different results. In general, the cognition
leads to simplified pastoral conclusion that the point density is
critical for accurate orientations.
If the differences are examined graphically (Figures 8-11), some
wave-like behaviour is found in all inspected directions. Likely,
this phenomenon is caused mainly by the small inaccuracies
with GPS and INS combined to fluctuation of the aeroplane.
Beforehand, also some systematic rotation between laser strips
was expected. However, visual study (Figures 8-11) did not
reveal any clear rotations. If necessary, the rotation parameters
could have been calculated using solved differences from test
sites as corresponding points in the last squares adjustment.
6. CONCLUSIONS
The repeatability of the laser measurements was investigated
using five almost completely overlapping laser strips measured
with TopoSys Falcon. The differences between strips were
measured in thirty-nine small test sites from the test area
covering 1500x100 meters. One strip was selected as a
reference strip and four others were compared to that one. In
each test site the entity of two laser point clouds were oriented
directly to the same coordinate system using interactive
orientation method.
The repeatability of elevations, according the test sites, was
excellent. The largest systematic bias was -0.014 m. With other
strips no significant systematic bias was found. In addition, the
standard deviation was 0.011, or less, for every comparison
confirming the homogeneity of elevation measurements. Even
maximum differences were only 0.02-0.04 m depending on the
strip. The flight direction did not make any noticeable
difference to repeatability.
The planimetric repeatability was not as good as with heights.
However, the maximum systematic biases of 0.064 meters in
along-track direction and -0.019 meters in across-track direction
are still quite reasonable. The bias and deviation in across-track
direction may have been underestimated, because there were
less suitable tie features for that direction and because of the
properties of TopoSys Falcon scanning footprint (Figure 1).
The flight direction was the most distinguishable reason of
systematic planimetric errors. When the strips, flown from the
same direction, are compared among each other, the maximum
bias was only -0.014 m in the along-track direction and 0.006 m
in across-track direction.
Some non-systematic errors were found within the laser strips.
Typically, these errors were accumulated making wave-like
pattern, leading to the conclusions the main source of these
errors is inaccuracies of GPS and INS. Against the assumptions,
there were no clear differences, whether the test area located in
the middle or in the either side of the strip. Obviously, the
system calibration has been sucessed well with TopoSys Falcon.
The laser strips are not completely homogenous. The
repeatability in altitudes is excellent, but the planimetric
variations slightly reduce the usability of this information.
Therefore, the main concern when improving the quality of
laser data is, how to get the planimetric accuracy into as
uniform quality as possible.
7. REFERENCES
Ahokas, E., Hyyppé, J., Kaartinen, H., 2004. A quality
assessment of repeated airborne laser scanner observations. In:
International Archives of Photogrammetry and Remote Sensing,
2004, Istanbul.
Burman, H., 2000. Adjustment of laser scanner data for
correction of orientation errors. In: International Archives of
Photogrammetry and Remote Sensing, vol. 33, part B3/1, pp.
119-126.
Burman, H. 2002. Laser strip adjustment for data calibration
and verification. In: /SPRS Commission III, Vol. 34, Part 3B
