[magick-users] Pinbarrel distortion correction

[Fred Weinhaus]

I believe that the easiest way, perhaps I am wrong, would simply be 
to photograph a true grid of a reasonable card size such as neutral 
gray card sizes, perhaps a foot on a side. Then simply do a least 
squares fit to the Dersch polynomial in R in order to compute the 
coefficients. There are programs that are available to help pick 
control points.  The one I know of is a plug in to ImageJ.

See
http://rsb.info.nih.gov/ij/
http://rsb.info.nih.gov/ij/plugins/
http://bigwww.epfl.ch/thevenaz/pointpicker/

The least square fit probably could be developed rather easily by 
some programmer to be fed from this or some similar file and could be 
offline from IM. Perhaps even Excel might be able to do it, but I am 
not an expert using Excel.

Fred


>Anthony and Fred,
>
>I have tested a few tools that claimed to determine the correct
>distortion correction coefficients for a certain camera type, but could
>not find one that did the job reliably. Some of them used specific test
>images of a special target (like equally spaced lines) that had to be
>photographed.
>
>This really complicates the process of calibrating a certain camera
>type, as these target have to be quite large in order to be photographed
>from some distance. Thus I think that the approach taken by PTlens is
>much more promising, i.e. using modern architecture buildings as
>ready-made targets. Anyway, I would like to know which programm you were
>referring to, Anthony, in order to give it a try.
>
>So far, I didn't know that this distortion has already been implemented
>in IM, so thanks for the hint. But as I said, as long as we do not have
>a reliable tool to determine the parameters of a certain camera type,
>this program option is rather academic.
>
>I will try to work out the approach in somewhat more detail, applying it
>to a real test photograph and then come back on the subject again. This
>will hopefully be a matter of a few days...
>
>Wolfgang
>