I have just worked out a stable and rather simple way to determine the
lens correction parameters. I uploaded the example image, the corrected
image an the Hugin project file to
www.unfallrekonstruktion.de/imagemagick/Olympus_C2500L.zip

The Olympus C2550L is an older prosumer zoom camera with a quite
pronounced barrel distortion at shorter focal lengths -- which are now
almost perfectly corrected.

The basic idea is to use modern faccades as targets, which is in fact
not my own idea, but that of Tom Niemann, author of PTlens, see
http://epaperpress.com/ptlens

In my examples, the faccade consists of square "tiles", such that we can
use the corner points of the "tiles" as targets. I chose 32 target
points, i.e. 4 horizontal lines with 8 points each. I then determined
the coordinates of these points by the help of WinMorph
(www.debugmode.com/winmorph), which relieves me from the workload of
having to pick every point twice, like in Hugin.

I then constructed the point list for the PTO file (by the help of
Excel), combining the 32 source points to 4 horizontal and 8 vertical
lines. For the target points, I just reversed the ordering of the
horizontal points, as I found that if the source and target points are
totally identical, the optimiser just doesn't work (?).

The results for the parameters a, b, c yielded by Hugin's optimiser were
then used to perform a distortion correction with ImageMagick (see
http://www.imagemagick.org/Usage/windows/#vb_example and
http://www.imagemagick.org/Usage/distorts/#barrel), which turned out to
be very good.

So I basically suggest

1) to use modern faccades as suitable targets
2) to generate the point list for the optimisation not with Hugin, as
this generates a needless overhead.

I would also suggest to rewrite the Simple Lens Calibration example in
regard to this.

I am however not sure

1) whether the ordering of the horizontal points really has to be
reversed for the target image

2) how to initiate the optimisation without using Hugin, which now more
or less just creates a superfluous overhead.

3) whether it would be more efficient to use horizontal and vertical
lines instead of general lines -- the horizontal lines however being not
strictly horizontal in the image, due to rotation of the camera.

Could you help me in this regard?

I'm planning to check the results of my manual optimisation against
those achieved by the calibrate_lens program, thus I'm looking forward
to the Windows EXE file of that program.

Wolfgang Hugemann

-- 
You received this message because you are subscribed to the Google Groups 
"hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Reply via email to