GEOMETRIC STABILITY OF LOW-COST DIGITAL CONSUMER CAMERAS
Thomas Läbe and Wolfgang Förstner
Institute for Photogrammetry, University of Bonn, Nussallee 15, 53115 Bonn, Germany
laebe@ipb.uni-bonn.de, wf@ipb.uni-bonn.de
Working Group 1/2
KEY WORDS: Camera calibration, stability of interior orientation, effect of zoom, nominal camera parameters.
ABSTRACT
During the last years the number of available low-cost digital consumer cameras has significantly increased while their
prices decrease. Therefore for many applications with no high-end accuracy requirements it is an important consider-
ation whether to use low-cost cameras. This paper investigates in the use of consumer cameras for photogrammetric
measurements and vision systems.
An important aspect of the suitability of these cameras is their geometric stability. Two aspects should be considered: The
change of calibration parameters when using the camera's features such as zoom or auto focus and the time invariance of
the calibration parameters. Therefore laboratory calibrations of different cameras have been carried out at different times.
The resulting calibration parameters, especially the principal distance and the principal point, and their accuracies are
given. The usefulness of the information given in the image header, especially the focal length, is compared to the results
of the calibration.
1 INTRODUCTION
During the last years the number of available low-cost dig-
ital consumer cameras has significantly increased while
their prices decrease: digital cameras with 2000 x 3000
pixels cost below 1000 Euro, while the additional price for
cameras in mobile phones having up to 640 x 480 Pixels
is significantly below 100 Euro, similar to web cameras.
Parallel to these hardware developments tools for geomet-
ric processing digital images become part of the software
delivered with the cameras: prominent examples are tools
for warping images, e. g. for morphing or for generating
plane or cylindrical panoramas.
Photogrammetry has to face the fact: all basic image pro-
cessing techniques are already in the hands of students
when they start learning. Though the awareness of lens dis-
tortion, especially of short focal length cameras is present,
geometric precision seems to be of no concern.
Aiming at ultimate precision is a classical goal in geodetic
and photogrammetric applications for two reasons: effi-
ciency, by exploiting the potential of the measuring device
‘image’, and the statistical properties of optimal solutions
are well understood, easing the evaluation with respect to
all kinds of errors in the assumed models. Close range
photogrammetry partly follows this paradigm.
However, accuracy requirements vary tremendously: from
relative accuracies 1 : 200, e.g. when measuring sizes of
windows at facades to 1 : 100 000 e.g. when determining
forms of airplane wings. In order to exploit the full po-
tential of photogrammetric techniques, especially in those
application areas with no high-end accuracy requirements
it is important whether use low-cost cameras is feasible.
There is a large body of literature on the calibration of
cameras in general, cf. the review in (Fraser, 1992). A
historical review on the development of consumer cameras
is given in (Clarke and Fryer, 1998). However, only few
investigations of low-cost cameras exist, e.g. (Kunii and
Chikatsu, 2001), however not treating the stability. The in-
terior orientation of zoom cameras has been investigated,
e.g. in (Burner, 1995) and (Wiley and Wong, 1995), e.g.
demonstrating the principal point not varying linearly with
zoom. On of the few investigations into the stability of
cameras over time is (Peipe and Stephani, 2003), however,
for a camera which is partly designed for the use in pho-
togrammetry, and therefore turned out to be quite stable,
cf. also (Shortis and Beyer, 1997).
In the computer vision area many of the classical models
for calibration, e.g. (Brown, 1966), (Brown, 1971) have
been adapted, cf. (Hartley and Zisserman, 2000). How-
ever, the dramatic changes of interior orientation when ex-
ploiting zoom cameras lead to a different view on the role
of calibration. First, there are two sets of parameters of
interior orientation (1) those five, which still guarantee a
straight line preserving mapping and (2) all others, which
lead to non-linear distortions. Second, starting from im-
ages with a straight line preserving camera model, the 3D-
reconstruction is performed in a projective coordinate sys-
tem, including the self-calibration of the five parameters of
the interior orientation of all images, and in a second step,
called stratification, an absolute orientation using control
points is performed to obtain an Euclidean 3D-reconstruction.
The loss in precision, especially due to the assumption of
a totally varying interior orientation, usually is accepted.
However, methods - similar to the plumb-line method of
D. C. Brown (Brown, 1971) have been developed to correct
only for the non-linear distortions (Devernay and Faugeras,
2001). No information on the time-stability of these non-
linear distortions appears to be available.
This paper investigates in the use of consumer cameras for
photogrammetric measurements and vision systems. An
important aspect of the suitability of these cameras is their
geometric stability. Two aspects should be considered:
The change of calibration parameters when using the cam-
era's features such as zoom or auto focus and the time in-
[nternatic
Table |
costs o
varianc
tory ca
at diffe
peciall
their a
tion gi
is com
The pa
camerz
measul
age di:
for the
tion of
change
The pa
achieve
and rec
2 SE
21 D
We sele
the spe«
to selec
guishec
Two we
ITECH |
only all
x 480
compre
settings
pressioi
VGA re
sion rat
with an
Two di;
The HE
ever a r
investig
with 3
pressior
features
zoom.
resoluti
pixels.
