sir
u
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004
| Camera || Chip size (pixel) | lens | price (Euro) |
Logitech Quick Cam Zoom 640x480 manual focus, fixed zoom 64
Terratec 2move 1.3 1280x1024 fixed focus, fixed zoom 79
HP Photosmart 435 2069x1560 fixed focus, fixed zoom 139
Sony DSC V1 2592x1944 autom. + manual focus, optical zoom 4x 550
Kodak DSC 460 3060x2036 autom. + manual focus, 24mm lens 1000-5000
Table 1: Data of the different cameras used in this study. Note that the Kodak DCS 460 is not available any more. The
costs of comparable cameras may be in the given range (1000-5000 Euro).
variance of the calibration parameters. Therefore labora-
tory calibrations of different cameras have been carried out
at different times. The resulting calibration parameters, es-
pecially the principal distance and the principal point, and
their accuracies are given. The usefulness of the informa-
tion given in the image header, especially the focal length,
is compared to the results.
The paper is organized as follows: After characterizing the
cameras selected for our investigation and the achievable
measuring accuracy we discuss temporal variations of im-
age distortions modeled by various additional parameters
for the interior orientation, the effect of changing resolu-
tion of the images, the effect of zooming, the effect of
change in focus and the effect of changing the aperture.
The paper closes with a short comparison with the results
achieved with the calibration tool contained in MATLAB
and recommendations.
2 SETUP OF INVESTIGATION
2.1 Description of the Cameras
We selected five cameras for our investigation which cover
the spectrum of consumer cameras. As most cameras allow
to select the resolution or the compression rate we distin-
guished altogether 10 different cameras’.
Two web cams represent the low end. The web cam LOG-
ITECH QUICK CAM ZOOM is a fixed zoom camera which
only allows to change focus. Only VGA modus with 640
x 480 pixels is investigated, with the implemented low
compression rate. The TERRATEC 2MOVE 1.3 has fixed
settings, but a larger sensor, compensated by a strong com-
pression ratio. We investigated the camera in two versions:
VGA resolution with 640 x 480 pixels, with a compres-
sion ratio of 11.3 and the full resolution with 1 Mega-pixel
with an even higher compression ratio of 15.9.
Two digital cameras represent the moderate price range.
The HP PHOTOSMART 435 also has fixed setting, how-
ever a much larger sensor than the Terratec web cam. We
investigated the VGA mode and the high resolution mode
with 3 Mega-pixels with the '"better' and 'optimal' com-
pression. The SONY DSC V1 has been selected due to its
features, namely automatic and manual focus and optical
zoom. Also here we investigate the VGA mode and two
resolutions, namely with 1 Mega-pixels and with 5 Mega-
pixels.
For comparison, the well known Kodak DCS 460, a 6 Mega-
pixel-camera, has been investigated. Here no compression
takes place. The used lens has a focal length of 24 mm.
The naming of the 10 different 'cameras' and their basic
features are given in the table
[ Code | Camera | format/size | compr. ratio |
1 Logitech VGA 29
2V Terratec VGA 11.3
2M 1 Mbyte 15.9
3V HP VGA 5:2
3MB 3 Mbyte "better': 6.7
3MO 3Mbyte | 'optimal': 5.1
4V Sony VGA 22
4_1M 1 Mbyte 25
4M 5 Mbyte 2.8
S Kodak 6 Mbyte -
Table 2: Code of investigated cameras. The compression
ratios are determined from the file sizes.
All cameras have the possibility to work with a fixed focus
setting.
2.2 Description of the Calibration Process
We used our tool TCC for calibrating digital cameras, cf.
(Abraham and Hau, 1997),(Abraham and Forstner, 1997),
(Abraham, 2000), consisting of a 3D-test field (cf. figure
1) and software for self-calibration.
The automatic detection and location of the points of the
3D-test field lead to image coordinates, which in a self-
calibrating bundle adjustment with blunder detection and
elimination lead to estimated values of the exterior and in-
terior orientation and coordinates of the 3D points, which
a priori are only known approximately. Details are given
in (Abraham and Hau, 1997).
The calibration of all cameras was performed as a test field
calibration. In all cases 24 images were taken in different
orientations in order to capture all types of distortions. All
camera parameters, such as zoom, focus, aperture, white
balance, etc. were kept constant during data capture as far
as possible.
All calibrations were performed indoor. The calibrations
were performed 3 to 7 times within a time period of four
months, always with the same calibration setup.
In all cases we have the principal distance, the coordinates
of the principal point and the scale difference in x’ and y’-
529
