[alma-config] image library

[John Conway]

On Thu, 13 Apr 2000, Min Yun wrote:

> 
> This is a reminder for everyone to vote on your 5 favorate images
> in the image library.  Please vote by this weekend, and I will announce
> the final results next week.  Again, the candidate images are found at:
> 
> 	www.aoc.nrao.edu/~myun/mma/imaging/sil.html
> 	

Some nice images have been collected and it has been hard to choose;
I'm not sure I have made the right choices, but I guess that
we should  choose 5 sources and start simulations.

I have had in mind  some of the practical difficulties in 
converting these input images into useful input FITS images, which
I found in making my earlier test images in my memos. If you
generate uv data direct from the images using FFT then you can think
that each image plays the role of the true sky  brightness distribution,
per pixel area, for CLEAN deconvolutions  you also need a lower resolution
comparison image with FWHM equal to the restoring beam (also for MEM
simulations with definite resolution, which is  the correct approach I
think; asking what the sky is like at a given  resolution is a well posed
question while if no resolution is specified I don't believe the question
is well posed).

I found that to achieve high DR and avoid aliasing needs 4 (and
preferably 5) pixels per beam FWHM of the fixed resolution comparison
image, hence in order to get sufficiently complex test images quite large
input images are needed. In addition to this some images will have to be 
modified to reduce  the large values which cut across the edges of the
images (e.g the filamaentary images), and finally it would be advantagous 
to modify  the image sizes to have 2^{N} on a side. 

           John.


  
Anyway my 5 votes go to; 

Cygnus A   - hard to choosewith Cas A, both similar very high 
              dynamic range. I choose the Cynus A image because the 
               Casa A image is actuall relatively small (248x249 pixels in
              the .gif), even if you choose a 1 lobe version of the Cyg A
                test  image the size is bigger by a factor of two. 
             

M51-Halpha   - More complex than the m51CO, test set should 
               contain a galaxy for promotional reasons.


Crab -   The Crab, G23, G111 and Oph all have similar filamentary
             structures. The G111 image looks complex but is only 
             415 x 199 pixels, and has its brightest emisison toward thee
             edge
            
            The  Crab has 760x742 pixels and has  the highest 
            DR (500) of the filamentary images, can be a good proxy for 
            similar structures in CO and  dust(?).  To avoid 'edge 
            effects' would still need to massage slightly.

Mars -  I think its nice to have as many quite different test images  as 
        possible. Probaly subtract a uniform disk model from image
          to make final test image, because this is how it would likely be 
           mapped.

10KDOTS - Again a different imaging challenge. Has an assocayed CC 
            extension, so the test data  be created directly from that
              may avoid limited dynamic range form image FFTs
              May be working making a bigger version of the same image
                and using that.