RE: Minimum spanning circles and Voronoi diagrams

[Tom Droege <tdroege2@earthlink.net>]

I cannot imagine that one can do as well "deblending" as one does with a 
clean image.  So I would want to mark such measurements as "not first quality".

OK the least I would do is to keep separate lists.  Or mark the stars as 
"clean", "deblended", etc..  We have lots of stars.  I lean toward a list 
of very clean stars for a start.  Well, a combined list with the proper 
label is just fine.  I think each measurement has to include several 
quality markers.  Some that I can think of:  Seeing, Pixel Problems, Cosmic 
Ray problems, ...  I am not sure we can do this with one byte.  Probably a 
half dozen bytes will be needed to give various quality levels for each 
item like clouds, seeing, nearby track, etc..

Note that the scheme that I have proposed is designed to detect problems 
due to bad pixels, cosmic rays, nearby track,  and blended stars.  Anything 
that "moves" a star.  It is going to be tough enough studying measurements 
which are as clean as possible.  Measurements that are not "clean" just 
produce outliers that make it difficult to see what is going on.  The 
problem as I see it is to cut the data set without introducing a 
bias.  Seems to me that an astrometry cut does this very well.  But so far 
this is a gedanken experiment.  I would like to turn it into a real one.

Most of the other folks doing this seem to be looking at dense star fields 
(STARE is an example - they have 40,000 stars in their typical field.  I 
find 2-3000 away from the galactic equator.)  This is the usual greed of 
science.  ;^)  They must have a much larger problem with blends than we 
do.  So we should get better results.  I am looking uniformly around a band 
at 7.5N.  Just a few of the fields are crowded.  Most people don't seem to 
look at uncrowded fields.  So I am interested in doing as good a job as is 
possible where people usually do not look.

Tom Droege







Tom Droege

At 04:09 PM 2/6/02 -0800, you wrote:

>--- Tom Droege <tdroege2@earthlink.net> wrote:
>
>
> > ...go through the real data.  Try to fit the psf.
> > Note that this is we are doing in the first pass star
> > finding.  At this point, I would
> > just
> > throw out the stars that do not fit the psf well.  This would
> > indicate
> > blended stars and the like.
>
>No, that's one of the best features of PSF fitting.  You can
>de-blend stars.  First you subtract a PSF from the blend
>and then you try and fit a PSF to the remainder.  I think
>daophot tries hard to do this using various combinations of
>fits until PSF subtraction yeilds some kind of minima.
>
>
>=====
>Chris Albertson
>   Home:   310-376-1029  chrisalbertson90278@yahoo.com
>   Cell:   310-990-7550
>   Office: 310-336-5189  Christopher.J.Albertson@aero.org
>
>__________________________________________________
>Do You Yahoo!?
>Send FREE Valentine eCards with Yahoo! Greetings!
>http://greetings.yahoo.com