Re: [APML]: Image Registration

["Wil Milan" <wmilan@airdigital.com>]

The Astro-Photography Mailing List
------------------------------------

I've had the same problem with Picture Window, and that was with virtually
identical photos, taken back to back with the center and orientation exactly
the same. (The autoguider continued uninterrupted during the two exposures,
so these two photos were as exactly aligned as is humanly possible.)
Nevertheless, a 4-point alignment near the corners still left very visible
misalignments over most of the photo. I can't explain it, but I'm glad to
hear I'm not the only one struggling with it. Any ideas?

Wil M.


-----Original Message-----
From: Chuck Vaughn <aa6g@aa6g.org>
To: astro-photo@nightsky.com <astro-photo@nightsky.com>
Date: Wednesday, September 23, 1998 2:22 PM
Subject: [APML]: Image Registration


>The Astro-Photography Mailing List
>------------------------------------
>
>Most of you have seen the tricolor images I've worked on over
>the past few months. Some are mine and some are done by my
>friend Robert Bickel. I've run into some interesting alignment
>problems that are not solved by any existing software I know of.
>
>The worst problems have been with Robert's images. He uses a
>8"-f/6 Newtonian with custom coma corrector. It is not possible
>to align all the stars in the field for all three images unless
>the center of the fields are almost exactly the same - easier
>said than done.
>
>Robert's system introduces a distortion that is absent or much
>reduced in my refractor. Aligning the image in the center by
>using four points in the corners produces a misalignment that
>roughly looks like this: A small circle in the center of good
>alignment, a donut around that where all the stars need to be
>moved towards the center, a larger donut of good alignment and
>finally the remaining outer portions and corners where the stars
>need to be moved out. This is totally non-correctable with only
>four points.
>
>A mathematical solution to replot the data would be best but I
>don't know what the formulas might be to account for this kind
>of distortion. Perhaps a more practical and straightforward solution
>would be software that allowed you to move and then fix as many
>points as required to get alignment. For example, align a star in
>the middle and fix that point. Move from the center, pick another
>star and align that so the stars from the center to the new point are
>aligned. Fix that new point. Repeat until you've worked your way to
>the edges and corners. Yes it's slow but it would work and it
>could be quite accurate.
>
>Do any of you who have been in contact with the PW author think he
>might be interested in adding something like this? I suspect the
>chances of ever getting a PS plug-in to this are rather slim.
>
>Chuck <aa6g@aa6g.org>
>