688
September 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
Q: I have read many papers regarding self-calibration
parameters usually found in aerial triangulation software,
but I don’t understand the full picture. The more I read,
the more they become too mathematical. When I work
on aerial triangulation tasks using INPHO, there are two
self-calibration parameters groups, namely the 12 param-
eters and 44 parameters. I mainly use 44 parameters to
finalize my projects. Using the self-calibration option
eliminates most—if not all—of the errors existing around
the GCPs and checkpoints. We always check our projects
quality in stereo environment and most the time we found
that products for projects adjusted with self-calibration
meet the accuracy standard while projects adjusted
without self-calibration has floating and digging problems
on the control and checkpoints. Does this mean that
our camera has a distortion even if it is not stated in the
calibration file? We use an Ultracam Eagle, which is a
fairly new sensor, and according to the manufacturer,
the residual distortion is less than two microns and it
does not need to be used during the aerial triangulation
process. What are these parameters and how do they
work? How reliable and acceptable are self-calibration
parameters in the science of photogrammetry and in
the geospatial industry, and how do we analyze the
self-calibration results?
Mulualem Yeshitila, Ethiopia
Dr. Abdullah:
The concept of analytical camera self-calibration
was introduced to the mapping industry shortly after the
birth of analytical aerial triangulation in the 1960s. The late
Duane Brown pioneered the bundle adjustment with additional
parameters (AP) shortly thereafter. The concept was originated
within the close-range photogrammetry practice, where many
of the cameras used were not metric, and established analogue
laboratory calibration chambers such as the one developed for
aerial film cameras were absent. However, due to its benefits to
the user, the concept of self-calibration found its way to the aerial
photogrammetry practice. The manual of photogrammetry
describes two main goals that can be achieved by performing
camera system self-calibration. Those are:
1) Physical Correction Model:
This model provides
comprehensive photogrammetric calibration of the camera
in lieu of the laboratory camera calibration, where the
interior calibration parameters such as focal length,
principal point and lens distortion of the camera system
are determined or adjusted to certain accuracy. In addition,
this type of self-calibration uses additional parameters to
model the image coordinate distortion that is caused by
symmetric radial distortion, decentering lens distortion,
image plane unflattens and in-plane image distortion
1
.
An additional parameters model, also called the “Physical
Additional Parameters Model” is used to differentiate it from
the empirical model (to be discussed later). The well-known
eight –parameter model, which was originally developed for
film cameras decades ago, is found to be equally effective to
model physical distortions in modern digital cameras. This
model, not only models lens distortion if it exists, but can also
model several imperfections in the CCD array electronics
and the digital signal transmission, which sometimes results
in differential scaling between x and y coordinates. This type
of self-calibration is most effective in applications where the
exterior orientation parameters are to be used directly in the
mapping process such as digital ortho-rectification.
1
For definitions of these terms refer to the “Manual of Photogrammetry”, sixth edition,
American Society of Photogrammetry and Remote Sensing.
“all cameras, including the large
format metric digital cameras, inherit
systematic errors due to film or
CCD deformation and other physical
characteristics that cause deviation from
perfect satisfaction of the collinearity
condition”
Photogrammetric Engineering & Remote Sensing
Vol. 81, No. 9, September 2015, pp. 688–690.
0099-1112/15/688–690
© 2015 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.81.9.688
