RE: Focus indication

["Creager, Robert S" <CreagRS@LOUISVILLE.STORTEK.COM>]

Hey all,

Well, I'm almost done with a pass of the focus indication, but I'm stuck on:

>         and then the half-intensity points from that centroid
>         by interpolation.  This centroiding method is described

I have the centroid, but am unsure on the interpolation from the centroid to
calculate the FWHM.  Theoretically, I believe I should fit a gaussian to the
marginal sums (x and y, from star pixels only).  Is this the only way, the
wrong way, or is there a simpler way?  Strongly looking for the simpler
way...

Thanks,
Rob

> -----Original Message-----
> From: aah@nofs.navy.mil [mailto:aah@nofs.navy.mil]
> Sent: Wednesday, June 13, 2001 11:56 AM
> To: tass@listserv.wwa.com
> Subject: RE: Focus indication
> 
> 
> Well, my real question was:  when you start talking
> real-time, I assume you are adding features to Tom's
> observing code which is written in Qbasic.  Now I guess
> what you are doing is letting Tom spawn off to a C-code
> program similar to what you are doing with the fits
> file writer.
>   So some basic information:
>     (a) you do need to locate some star.  The obvious
>         way is for Tom to use the cursor and select
>         a star on some image display.  The less obvious
>         way is to find a star in the central 200x200 region.
>         That second method is not always easy; you have
>         to be able to reject saturated stars,
>         non-stars (such as cosmic rays or satellites),
>         and stars that are contaminated by neighbors.
>         Good luck -- this is the hard part.  The star you
>         want has peak intensity about half-well and no
>         neighbors within ~8 pixels in any direction.
>     (b) once a star is located, you form a 'box' around
>         the star, usually about 16x16 pixels.  In the
>         x direction, you add up all pixels inside the box
>         at each y position; that gives you a 16-element
>         vector, which is called the y marginal sum vector
>         (marginal comes from the 'margins' of the box).
>         Similarly, for the y direction, you add up all pixels
>         inside the box at each x position; this forms a
>         16-element vector that is orthogonal to the first
>         vector.  Once the vectors are formed, you can
>         find the star's centroid in each vector (direction),
>         and then the half-intensity points from that centroid
>         by interpolation.  This centroiding method is described
>         in any text that talks about centroiding, such as AIP
>         or Ron Stone's 1970's AJ centroiding article.  It is
>         quicker than fitting a full 2-D function and is adequate
>         for Tom's needs.
> Arne
>