Helen Burman
6 SUMMARY AND CONCLUSIONS
The two main reasons for developing the laser strip adjustment are alignment calibration and block adjustment to make
overlapping strips coincide in one surface. Three groups of observation equations are used in a least-squares adjustment
in order to make overlapping laser strips coincide. Two of them are based on finding features for matching elevation or
intensity values in plane and one is for finding corresponding surfaces for matching in height. Problems in forested
areas can be overcome by filtering vegetation from the laser data. The remaining ground surface can then be used for
matching.
In the alignment procedure, the main interest is to estimate the misalignment angles between the sensors. These angles
are the same for the whole flight. Therefore, one set of attitude shift parameters can be used for the whole block. The
estimates of shift parameters for the position are strongly correlated with the attitude errors. The estimate of the pitch
error is correlated with a shift along the flying direction and the roll error is correlated with a shift across the flying
direction. The position shifts has to be known or be the same for all strips if the misalignment is to be determined.
Misalignment in roll and pitch can be found from differences between two strips flown in opposite directions while at
least three strips in different directions are needed to solve all three misalignment angles. To get redundancy, the
recommended configuration for alignment is to cover the area from four different directions. The gradient matching
method assumes a continuos surface with only one elevation or intensity value for each pair of (X,Y) co-ordinates. In
this investigation, intensity gradients (e.g. between hardmade surfaces and grass) suited best for this. The reason for
this is probably that these occurred at flat surfaces not geometrically sensitive for scanning direction. Large gradients in
height are often found in buildings. They are not suited for matching as the roof reaches over the wall, which not
follows the criteria of a continuos surface. In addition to this, large height differences often produce shadows, disturbing
the surface reconstruction.
When the reference height grid is unknown, it is estimated before each iteration by calculating the mean height value of
all laser strips. The laser strips are corrected for orientation errors, which are updated from the last iteration.
As for alignment calibration, the method can be used for adjustment of larger blocks of overlapping laser strips. If the
accuracy in georeferencing does not match the precision of the laser scanning, there might be effects like multiple layers
in overlapping areas. This can be annoying in visualisation of the result and in visual or automatic interpretation
techniques. Some additional features should be added to the automatic matching procedure for adjustment of laser data.
One is modelling the intensity difference in one object between different laser strips. Another is including feature
extraction to match edges. A third is including the option of having the height model as an unknown in the adjustment.
Finally, some self-diagnosis should be included to assign weights for the observations. In the present version, weights
are based on the à priori variances of the observations. In this example, only shift of the attitudes and positions are used.
The method can be expanded to also model time dependent drift in the orientation parameters.
ACKNOWLEDGEMENTS
The Swedish Space Board is greatly acknowledged for their financial support of this work. Many thanks to TopEye AB
for their support and co-operation, providing all the data for the tests. Finally, many thanks to Prof. Kennert Torlegérd
at the Department of Geodesy and Photogrammetry at KTH for his support and guidance.
REFERENCES
Burman H. (2000): Calibration and orientation of airborne image and laser scanner data using GPS and INS. PhD
Thesis, TRITA-GEOFOTO 2000:11, KTH, Stockholm 2000.
Kilian J., N. Haala, M. English (1996): Capture and Evaluation of Airborne Laser Scanner Data. /APRS 31-B3, Vienna
1996.
Lemmens M.J.P.M. (1997): Accurate Height Information from Airborne Laser-Altimetry. /n proceedings from IGARSS
’97, ISBN 0-7803-3839-1, pp. 423-426.
132 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.