Re: [APML]: Image Registration

[Chuck Vaughn <aa6g@aa6g.org>]

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

Mike,

> 
>     I have discussed this problem several times with
> Picture Window author Jonathan Sachs.  We were exploring a
> process which would do the stretching automatically based
> on a subtraction convergence to black.  Nothing has come of
> it so far.

This is fine for identical images but will not work for tricolor.
Even disregarding the chromatic abberation problem with refractors,
small faint stars appear on the red negative that don't on the
blue therefore it is not possible to subtract to black.
 
>    If this is still not good enough, you can divide the image in
> three equal parts along the scan axis.  Do a two point composite
> as described above for each section and then recombine the images as a
> mosaic.  This can produce excellent results but takes a lot of effort.

This could be good for some instances but could not handle the
problem I ran into where a donut of stars need to move toward the
center.

I considered chopping the image up into pieces and aligning the
pieces but that could get complicated and it would probably leave
lines where the images were connected. I'm not too enthusiastic
about that method.

If we're going to talk to him about this, we need to suggest a
method that will handle any misaligment.

Chuck <aa6g@aa6g.org>