Retrieve projection values from matplotlib plot with astropy/fits data

Question

I have a need to get the converted x,y pixel values of a matplotlib projection. In particular this is an astropy world coordinate system conversion, from a fits datafile. The data file provides a header which provides the projection information, but I know of no one to use it directly without knowing information I do not have. Here's the current code:

image_detection = fits.open("hst_12311_08_wfc3_uvis_total_drz.fits")['SCI'].data

wlist = fits.open("hst_12311_08_wfc3_uvis_total_drz.fits")['SCI']

w = wcs.WCS(wlist.header)
mean, median, std = sigma_clipped_stats(image_detection, sigma=3.0)

iraffind = IRAFStarFinder(fwhm=3.0, threshold=5*std, exclude_border=True)
sources = iraffind(image_detection - median)

positions = (sources['xcentroid'], sources['ycentroid'])
apertures = CircularAperture(positions, r = 4.0)

fig = plt.figure(figsize=(8,8))
ax = fig.add_subplot(111, projection = w)
ax.imshow(transform(image_detection), cmap='gray_r', origin='lower')
# ax.colorbar()
apertures.plot(color='blue', lw=1.5, alpha=0.5)
plt.savefig("apertures.pdf")
ax.xlabel('Right Ascension')
ax.ylabel('Declination')
plt.show()

What I'm after is those position values, given in x,y, converted to the world coordinates, which are plotted by the projection. I've checked the astropy documentation on WCS and it's not clear how certain values, such as those to do with the origin are obtained. The fits file used is freely available from the Hubble Legacy Archive, though any x,y data should be technically suitable. The fits header contains all of the values for the conversion as mentioned, though I do not fully understand their usage. This is a bit of a long shot here I think, but thank you if you can help.

Answer 1

Once you have the WCS object then you can use two methods to transform from x,y->RA,Dec or the reverse. These are w.all_pix2world() and w.all_world2pix() respectively eg

from astropy.io import fits
from astropy import wcs

wlist = fits.open("hst_12311_08_wfc3_uvis_total_drz.fits")['SCI']
w = wcs.WCS(wlist.header)

# Convert chip center from pixels to RA, Dec
radec_coords = w.all_pix2world(3057, 3045, 1)
print("RA, Dec=", radec_coords[0], radec_coords[1])

# Convert a RA, Dec pair to x,y
coords = [(279.1017383673262, -23.90274423755417)] # CRVAL1, 2 from header
pix_coords = w.all_world2pix(coords, 1)
print("X,Y=", pix_coords[0][0], pix_coords[0][1])

This will produce output (rounding permitting...) of

 RA, Dec= 279.10174439, -23.90274424
 X,Y= 3057.5, 3045.0

where the RA, Dec are in degrees. It's best to use the all_pix2world rather than wcs_pix2world to apply all the transformations (the core WCS transform from pixels to RA,Dec given by the CDi_j matrix and CRPIXi/CRVALi , along with any -SIP polynomials) if there are likely to be image distortions in the image (normally the case for Hubble instruments)