[fitsbits] Question(s) regarding development of proprietary FITS manipulation software. . .

[Rob Seaman]

On Aug 27, 2007, at 8:32 AM, Michael Williams wrote:

> I have a client that is interested in opening and editing FITS  
> files.  We've been provided with nothing more than FITS example  
> files, a FITS specification, and a few line items like "the ability  
> to view a histogram".

Perhaps your issue is more with your client's fuzzy description of  
the problem?  Since you mention a histogram I might guess that the  
application is specific to pixel arrays rather than tables?

> Again, I am not a scientist.

Perhaps solving your client's problem requires one?

> It seems that my problem is the fact that I'm expecting FITS to  
> actually *fit* (pun intended) into some specific category with  
> absolutely specific syntactical and semantic requirements.

No, as you say, FITS won't do that.  If your application is specific  
to pixel arrays, however, the range of categories is much narrower.

> 2) What is the significance of *any* imagery in FITS?  Are the  
> images literally live pictures that were taken, or some arbitrary  
> graphical representation of data?

All computer representations are arbitrary.  A CCD exposure of the  
sky represented as FITS is, however, as "living" a picture as  
treasured family snapshots in an album or socially ruinous JPEGs on  
Facebook.

To attempt to answer the question I think you think you're asking,  
however, unlike a GIF or JPEG, a FITS pixel array contains no  
explicit lookup tables conveying a false color mapping and no  
explicit color model conveying "true" color.

> Should I allow sepia toning?

I wouldn't forbid it, but no, don't spend much effort supporting such  
a feature.

> Should I allow them to run photoshop filters on the pictures?

Does your client's application have anything to do with publication  
graphics?

> I mean, what, really is the usefulness of any manipulation on the  
> image and

Each pixel represents a measurement.  Each array represents  
gazillions of such measurements.  Measurements require calibration to  
be scientifically useful.  Calibrated data are compared with other  
calibrated data to deduce assertions pertaining to empirical reality.

I'm not being facetious in asking whether your clients care about  
empirical reality.  If not, perhaps they care about FITS so they can  
list FITS compliance as a feature on a shrink wrapped box?

> what are the most common techniques?

Coaddition - by all means, coaddition.  Oh - and centroiding!  Ooh!   
Principle components analysis - not common now, but worth a passel of  
pork belly futures, mark my words.

My daughter's camera has "face recognition" to decide what part of  
the picture to focus against.  If your clients are involved in  
planetary imaging this could aid in detecting ROIs resembling Elvis.   
(Hmmm - another VO web service to explore...)

> 4) What is the goal of scientists when it comes to examining  
> another's FITS files?

The same thing we do every night, Pinky - try to take over the world!

Rob