Attention heat map

[C Oda]

I'd love to use the Calibration data (Center, Size, Pixel Scale & Orientation) of publicly uploaded images to create a map of which areas of the sky have received the most attention, and with which accuracy :)

Has this ever been done?

[Dustin Lang]

I made that map once, years ago.  I'd be happy to do a database dump for you to explore!

The database structure is managed by Django, via the models here, eg, here's Calibration

https://github.com/dstndstn/astrometry.net/blob/main/net/models.py#L607

and the WCS values are all stored in the TanWCS table:

https://github.com/dstndstn/astrometry.net/blob/main/net/wcs.py#L6

For making this heat-map, a rough approach would be to use the "SkyLocation" data,

https://github.com/dstndstn/astrometry.net/blob/main/net/models.py#L860

which give the nearest power-of-two "HEALpix" pixel for each image.  (Here's where that gets populated:

https://github.com/dstndstn/astrometry.net/blob/main/net/process_submissions.py#L394 )

If you let me know which tables you want, I can do the db dump.

cheers,

--dustin

On Mon, Nov 15, 2021 at 6:58 AM C Oda <ArndtS...@freenet.de> wrote:

I'd love to use the Calibration data (Center, Size, Pixel Scale & Orientation) of publicly uploaded images to create a map of which areas of the sky have received the most attention, and with which accuracy :)

Has this ever been done?

--
You received this message because you are subscribed to the Google Groups "astrometry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to astrometry+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/5160e3f8-8f47-4e56-9361-a39869051b17n%40googlegroups.com.

[C Oda]

The TanWCS table seems to be the most informative, so I'll try to use that

In the meantime, I'll try to find out in which ways to map this :)

Thank you very much!

[Dustin Lang]

I think you'll want to map it out on many scales.  Wide-field images are really a different kind of observation than single-deep-sky-object observations.

One thing you could do would be to put, say, a 100-pixel grid on the image and push each of those points through the WCS transformation to get RA,Dec.  Then accumulate those into, say, a Hammer-Aitoff or other all-sky projection.  That would give you basically a pixel count histogram.

Here's a database dump of the public solved image WCS values:

http://broiler.astrometry.net/~dstn/temp/wcs.csv.gz

http://broiler.astrometry.net/~dstn/temp/wcs.fits.gz

The columns are

        ra,dec,rad = tan.get_center_radecradius()
        print(imid, ra, dec, rad, tan.get_pixscale(),
              tan.get_orientation(), tan.get_parity(),
              tan.crval1, tan.crval2, tan.crpix1, tan.crpix2,
              tan.cd11, tan.cd12, tan.cd21, tan.cd22, tan.imagew, tan.imageh)

where 'tan' is a TanWCS,

https://github.com/dstndstn/astrometry.net/blob/main/net/wcs.py#L6

(the one file is called "csv" but it's actually in space-separated format -- sorry)

cheers,

--dustin

To view this discussion on the web visit https://groups.google.com/d/msgid/astrometry/86404931-d7e8-4182-b99e-1e09e2053c8dn%40googlegroups.com.

[C Oda]

Here's my first attempt https://i.imgur.com/tTTjyQN.png !

I'll publish the code here: https://github.com/void4/astroheat

I'm pretty new to all this astro stuff, so more complex projections will take some time, I need to learn more about projections/mappings etc.

(I'm not even sure if this code is accurate but looking at the map the results make sense so far)

Clear Skies!

- coda

[C Oda]

Here's an even cooler looking heatmap using matplotlibs contourf() function!

https://raw.githubusercontent.com/void4/astroheat/main/contour_1637021811.png

Also using only 25000 of the 1899073 rows so far

Updated the repo with it. First, run main.py to get the old visualization but also export the grid data to grid.pickle. Then run display.py to produce this.