The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part BI. Beijing 2008
305
In the first part, the results for each strip over each individual
building point out to some learning (Figure 3). Redundancy
(observations) can enhance the results namely buildings with
many observations (points) give better results. Buildings with
very small or big stdv, in their individual solutions, give
extremely different weights from the mean stdv to the solution.
They should be avoided from the total solution setting up a
threshold for big and small buildings, so only the medium size
buildings remain in the solution. The following thresholds were
used: (LiDAR 500-2000 points) and (TINs 500-1500).
Although the algorithm is capable of detecting these types of
errors (o x , a y , a z ), some buildings with errors have been located.
Possible errors can originate due to the photogrammetric
restitution (e.g. building 12). Through data snooping procedure
LiDAR points, 4-5% outliers are removed and the parameters
are stabilized. It is illustrated in Figures 5, 6 e.g. for angle
omega. The boresight offset components cannot be detected
accurately in this type of data due to high correlation of
parameters and noise. Thus in case of a bigger offset, it could
be detected (a x , a y , a z > b x , b y , b z ).
In the second part, the results of the total solutions were
analyzed, indicating that the algorithm can absorb the existence
of at least 15% ‘problematic-out of threshold’ buildings. This
should be considered as a restriction of this algorithm. Through
many tests with differently distributed buildings, it can be
concluded that positions similar to the Gruber positions, widely
used in photogrammetry, (6-8 buildings) are the optimum
(Figure 4). It can be noticed that the shape of buildings don’t
affect the results. This type of distribution of the buildings has
been actually confirmed in LiDAR boresight misalignment
solution (Csanyi and Toth, 2007). Combinations of strips are
necessary: at least, 2 strips flying in opposite directions are
needed to recover the signs of the parameters. Also a 3rd strip,
in a crossing direction, is preferred for enhancing the
incompleteness of the parallel strips. This cross strip could
decrease the effects of possible systematic errors which could
arise from many sources e.g. different flying height between
strips.
6. CONCLUSIONS
The feasibility of using urban areas for boresight misalignment
has been investigated. The influence of the number and
distribution of the necessary ‘building-positions’ on boresight’s
misalignment parameter estimation is evaluated. Experiments
with various number and distribution of ‘building-positions’ are
presented, analyzed and evaluated through QA/QC statistical
tests.
Under operational circumstances, the real accurate values of the
boresight misalignment are never accurately known and could
only be estimated. Furthermore, boresight misalignment
parameters could change over a relatively short time period.
Therefore, having a mechanism to almost continuously check it
is a valuable tool. In other words, the detection of possible
changes in the values (in the remaining boresight misalignment)
through a QA/QC validation process, can assure a sustained
product’s quality. These algorithms can be considered as a good
and fast tool for estimating parameters and detecting any
changes.
It is noticeable that this algorithm in not restricted by detailed
restitution, as only the main skeleton of a building is needed.
This is different from some other algorithms where, for instance,
the availability of roof planes is a prerequisite. In this algorithm
buildings should include points within the threshold range, and
at least 2 strips flying in opposite directions are necessary;
obviously a good distribution of the buildings is the necessity to
reach an optimum.
REFERENCES
Baltsavias, E.P., (1999). Airborne laser scanning: basic
relations and formulas. ISPRS Journal of Photogrammetry &
Remote Sensing 54, pp. 199-214.
Bossier, J. D., and Toth, C., (1995). Accuracies Obtained by the
GPSVan™. GIS/LIS, Vol. 1, pp. 70-77.
Burman, H., (2000). Adjustment of Laser Scanner Data for
Correction of Orientation Errors. IAPRS, Vol. XXXIII, Part B3,
pp. 125-132.
Csanyi, N., and Toth, C., (2007). Improvement of LiDAR Data
Accuracy Using LiDAR-Specific Ground Targets. PE&RS,
Volume 73, Number 4.
El-Sheimy, N., Schwarz, K. P., Gravel, M., (1995). Mobile 3-D
Positioning Using GPS/INS/Video Cameras. Mobile Mapping
Symposium, OSU Center for Mapping, pp. 236-249.
Filin, S., (2001). Recovery of Systematic Biases in Laser
Altimeters Using Natural Surfaces. IAPRS, Vol. XXXIV, Part
3/W4, pp. 85-91.
Filin, S., (2003). Analysis and implementation of a laser strip
adjustment model. IAPRS, 34(3/W13), pp.65-70.
Friess, P., (2006). Toward a rigorous methodology for airborne
laser mapping. International Calibration and Orientation
Workshop EuroCOW 2006.
Habib, A., Bang, K.I., Shin S., Mitishita E., (2007). LiDAR
system self-calibration using planar patches from
photogrammetric data. MMT’07, 'The 5 th International
Symposium on Mobile Mapping Technology
Kilian, J., Haala, N., Englich, M., (1996). Capture and
Evaluation of Airborne Laser Scanner Data. IAPRS, Vol. XXXI,
Part B3, pp. 383-388.
Morin, K., and El-Sheimy, N., (2002). Post-mission adjustment
of airborne laser scanning data. Proceedings XXII FIG
International Congress, Washington DC, USA, CD ROM.
Pothou A., Karamitsos S., Georgopoulos A., and Kotsis I.,
(2006a). Performance evaluation for aerial images and airborne
Laser Altimerty dara registration procedures. ASPRS, Nevada.
Pothou A., Karamitsos S., Georgopoulos A., and Kotsis I.,
(2006b). Assessment and comparison of registration algorithms
between aerial images and laser point clouds. ISPRS, Com. 1,
Symposium: From sensor to imagery ’ WGI/2, Part A, France.