Hans-Gerd Maas
The latter figures give a more realistic indication of the achieved precision of the shift parameters. The standard devia-
tion in height direction suffers from planimetric effects; when restricting constrained matching to almost flat patches,
sub-centimeter precision can be obtained in height direction.
Figure 5: Trend in shift parameters X-coord vs. X-shift PTT TE,
over test area (left = be- ass
fore, right = after refined M |
analysis of determinability
5. Application fields
Recent developments in the field
of direct geo-coding show rather
high precision figures for the
determination of sensor position
and orientation parameters (e.g.
Cramer, 1999), which might lead
to the conclusion that a tool for
the measurement of planimetric
shifts between laserscanner strip
data obsolete. Nevertheless,
there remain a number of reasons
justifying the application of least ole
squares matching in airborne 03
laserscanning:
e LSM may be used as a tool
for the verification of the 1
proper function of the system. 215
e In certain regions or under
certain circumstances, proper
differential GPS service may
not be available.
X-coord vs. Y-shift X-coord vs. Y-shift
X coord vs Z-shift X coord vs Z-shift
0.05
e LSM may be used as a tool for the improvement of the geometric quality of data obtained from low-cost laserscanner
systems, which are not equipped with a full high-grade GPS/INS system.
e The accuracy of LSM exceeds the accuracy of direct geo-coding systems at least in the height direction.
e Besides malfunction of the GPS/INS system, significant systematic errors in laserscanner strip data may also be
caused by misalignments between the position/orientation determination system and the actual laserscanner (van
Noort, 1999), or by other components of a laserscanner system. In this context LSM remains an important tool in the
procedure of self-calibrating laserscanning strip adjustment.
Beyond this, the method can be used as a general tool for matching 2'/,-D point clouds. In (Postolov et al., 1999) a similar
approach has been used for matching laserscanner data to surfaces generated interactively from stereo imagery. With
building models in a proper representation, the technique can also be used for matching given models to laserscanner
data. Similarly, laserscanner data can be fitted to given building models, which can be used as ground control. The latter
forms an analogy to a technique for fitting digital terrain models to not explicitly identified ground control points
presented by (Ebner/Ohlhof, 1994).
6. Future work
The main goal of the development of the matching technique is the provision of a tool for the detection and measurement
of discrepancies between neighbouring strips of airborne laserscanner data. An optimised matching strategy as well as a
tool for self-calibrating laserscanner strip adjustment still have to be developed. Further options of improvement include:
e Partial results: So far, the matching tool is designed to determine either all three shift parameters or only the vertical
shift. The latter can be selected by the user or enforced automatically after the analysis of the design matrix. In all
cases of a singularity in one planimetric direction, the match is rejected, although the shift parameter perpendicular to
that direction is still determinable. This option has not been implemented yet, as the current analysis procedure is
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 553
