International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
ID Camera JOP ae Num be ji 6, (mmn) Similar
1 Nikon288616 Set 1 Set 2 0.00618 No
2 Nikon288894 Set 1 Set 2 0.00475
3 Nikon288895 Set 1 Set 2 0.00311
4 Nikon288896 Set 1 Set 2 0.01603
Table 7 — Stability Comparison of IOP sets for Nikon cameras
(Note: If co « 0.0021, IOP sets considered similar)
As mentioned earlier, the ROT method is used to evaluate the
similarity between the IOP sets. However, if the SPR method is
used, the results differ depending on the type of terrain chosen.
A comparison was done between a flat and hilly terrain using
just the SPR method for the Nikon288616 camera. Two extreme
object space configurations were used. The first object space
represented a hilly terrain with a height variation of + 800 m.
The second object space represented a flat terrain with a height
variation of £ 0 m. For the hilly terrain, the spatial offset
standard deviation was 0.00619 mm, which is close to that
estimated by the ROT method (0.00618 mm). These standard
deviation values are similar because a hilly terrain would
decouple any correlation between the IOP and EOP, thus
yielding a reliable evaluation of the degree of similarity
between the reconstructed bundles. On the other hand, using a
flat terrain, the standard deviation of the spatial offset from the
SPR procedure turned out to be 0.000087 mm (approximately
0.03 pixels), which indicates a good fit between the two bundles
at a flat object space. However, this is a very optimistic and
deceiving conclusion. In such a case, a flat terrain would lead to
high correlation between the IOP and EOP. Therefore, although
the two bundles are significantly different from each other, the
EOP will adapt to absorb the differences between the two IOP
sets to produce a good fit at the object space.
6. CONCLUSION AND RECOMMENDATIONS
The presented research outlined an efficient approach for
calibration, and a meaningful measure for evaluating stability of
off-the-shelf digital cameras. For calibration, an easy-to-
establish test field consisting of straight line features and
signalized points were used. Deviations from straightness in
image space straight lines were attributed to various distortion
parameters that were modelled using collinearity equations.
The two methods of evaluating camera stability, which
quantitatively determined the degree of similarity between
reconstructed bundles using two sets of IOP, were introduced.
The ROT method is a comparison confined to the image space
where the quality of the coincidence between conjugate light
rays within two reconstructed bundles sharing the same
perspective center is determined. The two bundles are allowed
to rotate relative to each other until the best coincidence is
achieved. The SPR method allows for spatial and rotational
offsets between the two bundles while observing their quality of
fit at a given object space. For both methods, the similarity
measure was characterized by the standard deviation of the
spatial offset between the two bundles. If the standard deviation
was within the image coordinate measurement accuracy range
(V4 to % pixel size) of the implemented camera, then the two
[OP sets were considered similar.
There were nine amateur and professional digital cameras tested
in the experiments. Each type of camera had different
characteristics and resolutions. The experimental analysis of the
cameras revealed that the IOP remained stable over the eight-
month period. The only exception was the stability of the Nikon
cameras. The Nikons were just turned off and on between
dataset acquisitions and this altered the IOP.
It should be noted that the calibration technique and stability
measures described in this paper are general enough that they
can be applied to digital as well as analogue cameras intended
for mapping applications. The proposed stability measures
would allow amateur users of digital cameras to evaluate their
IOP and stability. A possible future initiative could be directed
towards finding a way to compare more than two sets of IOP at
one time. Current research will test additional new cameras and
continue to focus on their short and long-term stability.
ACKNOWLEDGEMENT
This research work has been conducted under the auspices of
the GEOIDE Research Network through its financial support of
the project (ACQZHAB: SIACQO5). The authors would also
like to acknowledge Mr. Paul Mrstik from Mosaic Mapping
Systems Inc. for his help in establishing the calibration test field
and providing some of the cameras used to analyze stability.
7. REFERENCES
Brüuer-Burchardt, C., and K. Voss, 2001. A new algorithm to
correct fish-eye and strong wide-angle-lens-distortion from
single images, Proceedings of the 2001 International
Conference on Image Processing, October 7-10, 2001,
Thessaloniki, Greece, Volume 1, pp. 225-228.
Brown, D., 1971. Close range camera calibration, Journal of
Photogrammetric Engineering & Remote Sensing, 37 (8):
855-866.
Chen, S., and W. Tsai, 1991. A systematic approach to analytic
determination of camera parameters by line features,
Pattern Recognition, 23(8): 859-877.
Fraser, C., 1997. Digital camera self-calibration, ISPRS Journal
of Photogrammetry and Remote Sensing, 52(4): 149-159.
Fryer, J, 1996. Camera calibration. Close range
Photogrammetry and machine vision, (K.B. Atkinson,
editor), Whittles Publishing, Caithness, Scotland, pp. 156-
180.
Guogqing, Z., U. Ethrog, F. Wenhao, and Y. Baozong, 1998.
CCD camera calibration based on natural landmarks,
Pattern Recognition, 31(11): 1715-1724.
Habib, A., M. Morgan, and Y. Lee, 2002-a. Bundle Adjustment
with Self-Calibration using Straight Lines, Journal of
Photogrammetric Record, 17(100): 635-650.
Habib, A., S. Shin, and M. Morgan, 2002-b. New Approach for
Calibrating Off-The-Shelf Digital Cameras, ISPRS
Symposium of PCV'02 Photogrammetric Computer
Vision, Graz, Austria, 9-13 September, Unpaginated CD
Rom.
Habib, A., M. Morgan, 2004. Stability Analysis and Geometric
Calibration of off-the-shelf digital cameras, Accepted for
publication in the Journal of Photogrammetric Engineering
& Remote Sensing.
International
CEE
Habib, A., Gl
Historic
CIPA -
Cultural
Heuvel, F., 1
from co
Proceed
and Rer
(5W11):
Koch, K., 1€
in Lin
Berlin, ]
Prescott, B.,
Radial
Processi
