def __init__(self, *args, **kwargs):
self.verts = None
self.weights = None
if len(args) == 0:
return None
assert len(
args) == 6, "args = im, im_header, locx, locy, ap_type, npix"
im, im_header, locx, locy, ap_type, npix = args
if ap_type == 'circular':
radius = np.sqrt(npix / np.pi)
_ap = circular_aperture(im, locx, locy, radius)
elif ap_type == 'region':
radius = np.sqrt(npix / np.pi)
_ap = region_aperture(im, locx, locy, npix)
else:
assert False, "ap_type must be circular or aperture"
verts = _ap.verts
verts = pd.DataFrame(verts, columns=['x', 'y'])
# Convert x, y to sky coordinates
wcs = astropy.wcs.find_all_wcs(im_header, keysel=['binary'])[0]
verts['ra'], verts['dec'] = wcs.all_pix2world(verts.x, verts.y, 0)
self.verts = verts
self.weights = _ap.weights
self.npix = npix
self.ap_type = ap_type
self.name = "{}-{:.1f}".format(ap_type, npix)
def get_solid_angle(hdu):
wcs = astropy.wcs.WCS(hdu.header)
nx = hdu.header['NAXIS1']
ny = hdu.header['NAXIS2']
xx = numpy.arange(-0.5, nx, 1)
xy = numpy.arange(0, ny, 1)
yx = numpy.arange(0, nx, 1)
yy = numpy.arange(-0.5, ny, 1)
xX, xY = numpy.meshgrid(xx, xy)
yX, yY = numpy.meshgrid(yx, yy)
xXw, xYw = wcs.all_pix2world(xX, xY, 0)
yXw, yYw = wcs.all_pix2world(yX, yY, 0)
dXw = xXw[:, 1:] - xXw[:, :-1]
dYw = yYw[1:] - yYw[:-1]
A = abs(dXw * dYw) * deg * deg
return A
def get_solid_angle(hdu):
wcs = astropy.wcs.WCS(hdu.header)
nx = hdu.header['NAXIS1']
ny = hdu.header['NAXIS2']
xx = numpy.arange(-0.5, nx, 1)
xy = numpy.arange(0, ny, 1)
yx = numpy.arange(0, nx, 1)
yy = numpy.arange(-0.5, ny, 1)
xX, xY = numpy.meshgrid(xx, xy)
yX, yY = numpy.meshgrid(yx, yy)
xXw, xYw = wcs.all_pix2world(xX, xY, 0)
yXw, yYw = wcs.all_pix2world(yX, yY, 0)
dXw = xXw[:,1:] - xXw[:,:-1]
dYw = yYw[1:] - yYw[:-1]
A = abs(dXw * dYw) * deg * deg
return A
def pix2world(self, x, y):
""" Transform pixel coordinates to world coordinates.
Return a two-element tuple with the right ascension and declination to
which the specified x- and y-coordinates correspond in the FITS image.
Raises NoWCSInformationError if the header of the FITS image does not
contain an astrometric solution -- i.e., if the astropy.wcs.WCS class
is unable to recognize it as such. This is something that should very
rarely happen, and almost positively caused by non-standard systems or
FITS keywords.
"""
coords = (x, y)
wcs = self._get_wcs()
pixcrd = numpy.array([coords])
ra, dec = wcs.all_pix2world(pixcrd, 1)[0]
# We could use astropy.wcs.WCS.has_celestial for this, but as of today
# [Tue Jan 20 2015] it is only available in the development version of
# Astropy. Therefore, do a simple (but in theory enough) check: if the
# header does not contain an astrometric solution, WCS.all_pix2world()
# will not be able to transform the pixel coordinates, and therefore
# will return the same coordinates as the FITSImage.center attribute.
if (ra, dec) == coords:
msg = ("{0}: the header of the FITS image does not seem to "
"contain WCS information. You may want to make sure that "
"the image has been solved astrometrically, for example "
"with the 'astrometry' LEMON command.".format(self.path))
raise NoWCSInformationError(msg)
return ra, dec
def pix2world(self, x, y):
""" Transform pixel coordinates to world coordinates.
Return a two-element tuple with the right ascension and declination to
which the specified x- and y-coordinates correspond in the FITS image.
Raises NoWCSInformationError if the header of the FITS image does not
contain an astrometric solution -- i.e., if the astropy.wcs.WCS class
is unable to recognize it as such. This is something that should very
rarely happen, and almost positively caused by non-standard systems or
FITS keywords.
"""
coords = (x, y)
wcs = self._get_wcs()
pixcrd = numpy.array([coords])
ra, dec = wcs.all_pix2world(pixcrd, 1)[0]
# We could use astropy.wcs.WCS.has_celestial for this, but as of today
# [Tue Jan 20 2015] it is only available in the development version of
# Astropy. Therefore, do a simple (but in theory enough) check: if the
# header does not contain an astrometric solution, WCS.all_pix2world()
# will not be able to transform the pixel coordinates, and therefore
# will return the same coordinates as the FITSImage.center attribute.
if (ra, dec) == coords:
msg = ("{0}: the header of the FITS image does not seem to "
"contain WCS information. You may want to make sure that "
"the image has been solved astrometrically, for example "
"with the 'astrometry' LEMON command.".format(self.path))
raise NoWCSInformationError(msg)
return ra, dec
def __init__(self, fname, fill_value=0.0):
with fits.open(fname) as hdul:
rawdata = hdul[0].data
wcs = astropy.wcs.WCS(hdul[0].header)
# WL is in Angstroms -> to microns
all_wl = wcs.all_pix2world(range(hdul[0].data.size), 0)
self._interp = ii.interp1d(all_wl[0] / 1e4, numpy.abs(rawdata),
bounds_error=False,
fill_value=fill_value)
def pc_gaia_cat(exp, mag_thresh=None, edge_pad_pix=0, nmp=None,
max_n_stars=3000, pm_corr=False):
print('Reading Gaia DR2 catalogs...')
xgrid, ygrid = xy_subsamp_grid()
wcs = exp.wcs
# last arg is 0 rather than 1 because that's what agrees with IDL
ra, dec = wcs.all_pix2world(xgrid, ygrid, 0)
cat = read_gaia_cat(ra, dec, nmp=nmp)
if mag_thresh is not None:
keep = (cat['PHOT_G_MEAN_MAG'] <= mag_thresh)
assert(np.sum(keep) > 0)
cat = cat[keep]
if pm_corr:
gaia_pm_corr(cat, exp.header['MJD-OBS'])
x_gaia_guess, y_gaia_guess = wcs.all_world2pix(cat['RA'], cat['DEC'], 0)
par = common.pc_params()
keep = (x_gaia_guess > edge_pad_pix) & (y_gaia_guess > edge_pad_pix) & (x_gaia_guess < (par['nx'] - 1 - edge_pad_pix)) & (y_gaia_guess < (par['ny'] - 1 - edge_pad_pix))
assert(np.sum(keep) > 0)
cat = cat[keep]
x_gaia_guess = x_gaia_guess[keep]
y_gaia_guess = y_gaia_guess[keep]
cat = Table(cat)
cat['x_gaia_guess'] = x_gaia_guess
cat['y_gaia_guess'] = y_gaia_guess
if len(cat) > max_n_stars:
# retain brightest max_n_stars
# it'd be better to do this cut based on
# 'G_PRIME', the color-corrected pointing
# camera Gaia-based mag
# in the future can evaluate trying to spread the
# selected max_n_stars evenly across quadrants
# (could imagine a pathological case with e.g., a
# globular cluster in the FOV)
print('Restricting to the brightest ' + str(max_n_stars) + \
' of ' + str(len(cat)) + ' stars')
sind = np.argsort(cat['PHOT_G_MEAN_MAG'])
cat = cat[sind[0:max_n_stars]]
return cat
def indices_to_velocity(
indices: np.ndarray,
base_wavelength: u.Quantity,
wcs: astropy.wcs.WCS,
wscale: int
):
wavelengths, _, _ = wcs.all_pix2world(wscale * indices, 0, 0, 0)
wavelengths = wavelengths << u.AA
velocities = astropy.constants.c * (wavelengths - base_wavelength) / base_wavelength
velocities = velocities.to(u.km / u.s)
return velocities
def corner_catalog_1ext(telra, teldec, extname):
# LL -> UL -> UR -> LR
xpix, ypix = util.ci_corner_pixel_coords()
wcs = ci_wcs.nominal_tan_wcs(telra, teldec, extname)
ra, dec = wcs.all_pix2world(xpix, ypix, 0)
name = [(extname + '_' + corner) for corner in ['LL', 'UL', 'UR', 'LR']]
tab = Table([ra, dec, name], names=('ra', 'dec', 'name'))
return tab
def test_qphot_run_proper_motions(self):
# Do photometry on Barnard's Star, the star with the largest-known
# proper motion relative to the Solar System. Its coordinates are
# J2000, but the DSS image dates back from 1993: qphot.run() must
# correct the right ascension and declination before doing photometry,
# adjusting for the change in its position over those seven years.
barnard_path = "./test/test_data/fits/Barnard's_Star.fits"
barnard = astromatic.Coordinates(269.452075, 4.693391, -0.79858,
10.32812)
nbits = get_nbits()
if nbits == 64:
expected_output = ( # x y mag sum flux stdev
qphot.QPhotResult(440.947, 382.595, 17.245, 8563646, 5311413,
1039.844))
else:
assert nbits == 32
expected_output = ( # <<>>
qphot.QPhotResult(440.947, 382.595, 17.245, 8563646, 5312029,
1039.844))
path = fix_DSS_image(barnard_path)
with test.test_fitsimage.FITSImage(path) as img:
# Fix incorrect date in FITS header: '1993-07-26T04:87:00'.
# Subtract sixty minutes and add one hour: 4h87m == 5h27m
keyword = 'DATE-OBS'
assert img.read_keyword(keyword) == '1993-07-26T04:87:00'
img.update_keyword(keyword, '1993-07-26T05:27:00')
# The proper-motion corrected coordinates
year = img.year(exp_keyword='EXPOSURE')
expected_coordinates = barnard.get_exact_coordinates(year)
result = qphot.run(img, [barnard], **self.QPHOT_KWARGS)[0]
self.assertEqual(result, expected_output)
# Transform the pixel coordinates that IRAF's qphot outputs for
# each measured object to celestial coordinates. This allows us to
# make sure that photometry has been effectively done on the right,
# proper-motion corrected coordinates.
wcs = astropy.wcs.WCS(img._header)
ra, dec = wcs.all_pix2world(result.x, result.y, 1)
f = self.assertAlmostEqual
f(ra, expected_coordinates.ra, delta=1e-3) # delta = 0.24 arcsec
f(dec, expected_coordinates.dec, delta=1e-3) # delta = 3.6 arcsec
def test_qphot_run_proper_motions(self):
# Do photometry on Barnard's Star, the star with the largest-known
# proper motion relative to the Solar System. Its coordinates are
# J2000, but the DSS image dates back from 1993: qphot.run() must
# correct the right ascension and declination before doing photometry,
# adjusting for the change in its position over those seven years.
barnard_path = "./test/test_data/fits/Barnard's_Star.fits"
barnard = astromatic.Coordinates(269.452075, 4.693391, -0.79858, 10.32812)
nbits = methods.get_nbits()
if nbits == 64:
expected_output = ( # x y mag sum flux stdev
qphot.QPhotResult(440.947, 382.595, 17.245, 8563646, 5311413, 1039.844))
else:
assert nbits == 32
expected_output = ( # <<>>
qphot.QPhotResult(440.947, 382.595, 17.245, 8563646, 5312029, 1039.844))
path = fix_DSS_image(barnard_path)
with test.test_fitsimage.FITSImage(path) as img:
# Fix incorrect date in FITS header: '1993-07-26T04:87:00'.
# Subtract sixty minutes and add one hour: 4h87m == 5h27m
keyword = 'DATE-OBS'
assert img.read_keyword(keyword) == '1993-07-26T04:87:00'
img.update_keyword(keyword, '1993-07-26T05:27:00')
# The proper-motion corrected coordinates
year = img.year(exp_keyword = 'EXPOSURE')
expected_coordinates = barnard.get_exact_coordinates(year)
result = qphot.run(img, [barnard], **self.QPHOT_KWARGS)[0]
self.assertEqual(result, expected_output)
# Transform the pixel coordinates that IRAF's qphot outputs for
# each measured object to celestial coordinates. This allows us to
# make sure that photometry has been effectively done on the right,
# proper-motion corrected coordinates.
wcs = astropy.wcs.WCS(img._header)
ra, dec = wcs.all_pix2world(result.x, result.y, 1)
f = self.assertAlmostEqual
f(ra, expected_coordinates.ra, delta = 1e-3) # delta = 0.24 arcsec
f(dec, expected_coordinates.dec, delta = 1e-3) # delta = 3.6 arcsec
def pix2world(pix, header, axis):
wcs = astropy.wcs.WCS(header)
cunit = astropy.units.Unit(header.get('CUNIT%d' % (axis)))
pix = numpy.array(pix)
if pix.ndim == 0:
plen = 1
else:
plen = len(pix)
pass
p = numpy.zeros([plen, wcs.naxis])
p[:, axis - 1] = pix
world = wcs.all_pix2world(p, 0)[:, axis - 1] * cunit
if pix.ndim == 0:
return world[0]
return world
def convertImgCoords(coords, image, to_pix=None, to_radec=None):
''' Transform the WCS info in an image
Convert image pixel coordinates to RA/Dec based on
PNG image metadata or vice_versa
Parameters:
coords (tuple):
The input coordindates to transform
image (str):
The full path to the image
to_pix (bool):
Set to convert to pixel coordinates
to_radec (bool):
Set to convert to RA/Dec coordinates
Returns:
newcoords (tuple):
Tuple of either (x, y) pixel coordinates
or (RA, Dec) coordinates
'''
try:
wcs = getWCSFromPng(image)
except Exception as e:
raise MarvinError('Cannot get wcs info from image {0}: {1}'.format(
image, e))
if to_radec:
try:
newcoords = wcs.all_pix2world([coords], 1)[0]
except AttributeError as e:
raise MarvinError(
'Cannot convert coords to RA/Dec. No wcs! {0}'.format(e))
if to_pix:
try:
newcoords = wcs.all_world2pix([coords], 1)[0]
except AttributeError as e:
raise MarvinError(
'Cannot convert coords to image pixels. No wcs! {0}'.format(
e))
return newcoords
def pix2world(pix, header, axis):
wcs = astropy.wcs.WCS(header)
cunit = astropy.units.Unit(header.get('CUNIT%d'%(axis)))
pix = numpy.array(pix)
if pix.ndim == 0:
plen = 1
else:
plen = len(pix)
pass
p = numpy.zeros([plen, wcs.naxis])
p[:,axis-1] = pix
world = wcs.all_pix2world(p, 0)[:,axis-1] * cunit
if pix.ndim == 0:
return world[0]
return world
def center_wcs(self):
""" Return the world coordinates of the central pixel of the image.
Transform the pixel coordinates of the center of the image to world
coordinates. Return a two-element tuple with the right ascension and
declination. Most of the time the celestial coordinates of the field
center can be found in the FITS header, but (a) these keywords are
non-standard and (b) it would not be the first time that we come
across incorrect (or, at least, not as accurate as we would expect)
coordinates. Instead of blindly trusting the FITS header, compute these
values ourselves -- provided, of course, that our images are calibrated
astrometrically.
Raises NoWCSInformationError if the header of the FITS image does not
contain an astrometric solution -- i.e., if the astropy.wcs.WCS class
is unable to recognize it as such. This is something that should very
rarely happen, and almost positively caused by non-standard systems or
FITS keywords.
"""
center = tuple(self.center)
wcs = self._get_wcs()
pixcrd = numpy.array([center])
ra, dec = wcs.all_pix2world(pixcrd, 1)[0]
# We could use astropy.wcs.WCS.has_celestial for this, but as of today
# [Tue Jan 20 2015] it is only available in the development version of
# Astropy. Therefore, do a simple (but in theory enough) check: if the
# header does not contain an astrometric solution, WCS.all_pix2world()
# will not be able to transform the pixel coordinates, and therefore
# will return the same coordinates as the FITSImage.center attribute.
if (ra, dec) == center:
msg = ("{0}: the header of the FITS image does not seem to "
"contain WCS information. You may want to make sure that "
"the image has been solved astrometrically, for example "
"with the 'astrometry' LEMON command.".format(self.path))
raise NoWCSInformationError(msg)
return ra, dec
def get_coordinate_grid_gal(hdu, frame='fk5'):
"""Returns world coordinate grid of an image
[Image has to be in J2000 at the moment]
Gives both RA/DEC and GALACTIC
Input:
hdu = Input fits hdu in J2000 coordinates
frame = Input reference frame; default is fk5
Output:
(coordinate_grid_l, coordinate_grid_b) = coordinates of each array index in GALACTIC
(coordinate_grid_ra, coordinate_grid_dec) = coordinates of each array index in RA/DEC
"""
wcs = astropy.wcs.WCS(hdu)
shape = hdu.shape
index_grid_x = np.empty(shape)
index_grid_y = np.empty(shape)
coordinate_grid_ra = np.empty(shape)
coordinate_grid_dec = np.empty(shape)
coordinate_grid_l = np.empty(shape)
coordinate_grid_b = np.empty(shape)
for row in range(shape[1]):
for col in range(shape[0]):
index_grid_x[col, row] = row
index_grid_y[col, row] = col
coordinate_grid_radec = wcs.all_pix2world(index_grid_x, index_grid_y, 0)
coordinate_grid_ra, coordinate_grid_dec = coordinate_grid_radec
lb = coordinates.SkyCoord(coordinate_grid_ra,
coordinate_grid_dec,
frame=frame,
unit=(au.deg, au.deg)).galactic
coordinate_grid_l = lb.l.deg
coordinate_grid_b = lb.b.deg
return (coordinate_grid_l, coordinate_grid_b), (coordinate_grid_ra,
coordinate_grid_dec)
def update_flux_limits(header, pixlims, wcs=None, ref=1):
"""Update keywords used for flux limits"""
pixlim_coms = ["Start of valid flux calibration",
"End of valid flux calibration",
'Start of region with at least one fiber',
'End of region with at least one fiber',
'Start of region with all fibers',
'End of region with all fibers',
]
if ref not in [0, 1]:
raise ValueError("ref must be 0 or 1")
off = (ref + 1) % 2
if wcs is None:
wcs = astropy.wcs.WCS(header)
for key, com in zip(PIXLIM_KEYS, pixlim_coms):
header[key] = (pixlims[key] + off, com)
r1 = numpy.array([pixlims[key] for key in PIXLIM_KEYS])
r2 = numpy.zeros_like(r1)
lm = numpy.array([r1, r2])
# Values are 0-based
wavelen_ = wcs.all_pix2world(lm.T, ref)
if wcs.wcs.cunit[0] == u.dimensionless_unscaled:
# CUNIT is empty, assume Angstroms
wavelen = wavelen_[:, 0] * u.AA
else:
wavelen = wavelen_[:, 0] * wcs.wcs.cunit[0]
for idx, (key, com) in enumerate(zip(WAVLIM_KEYS, pixlim_coms)):
header[key] = (wavelen[idx].to(u.AA).value, com)
return header
def xy2rd(self, wcs, pixx, pixy):
""" Transform input pixel positions into sky positions in the WCS provided.
"""
return wcs.all_pix2world(pixx, pixy, 1)
def calcNewEdges(wcs, shape):
"""
This method will compute sky coordinates for all the pixels around
the edge of an image AFTER applying the geometry model.
Parameters
----------
wcs : obj
HSTWCS object for image
shape : tuple
numpy shape tuple for size of image
Returns
-------
border : arr
array which contains the new positions for
all pixels around the border of the edges in alpha,dec
"""
naxis1 = shape[1]
naxis2 = shape[0]
# build up arrays for pixel positions for the edges
# These arrays need to be: array([(x,y),(x1,y1),...])
numpix = naxis1*2 + naxis2*2
border = np.zeros(shape=(numpix,2),dtype=np.float64)
# Now determine the appropriate values for this array
# We also need to account for any subarray offsets
xmin = 1.
xmax = naxis1
ymin = 1.
ymax = naxis2
# Build range of pixel values for each side
# Add 1 to make them consistent with pixel numbering in IRAF
# Also include the LTV offsets to represent position in full chip
# since the model works relative to full chip positions.
xside = np.arange(naxis1) + xmin
yside = np.arange(naxis2) + ymin
#Now apply them to the array to generate the appropriate tuples
#bottom
_range0 = 0
_range1 = naxis1
border[_range0:_range1,0] = xside
border[_range0:_range1,1] = ymin
#top
_range0 = _range1
_range1 = _range0 + naxis1
border[_range0:_range1,0] = xside
border[_range0:_range1,1] = ymax
#left
_range0 = _range1
_range1 = _range0 + naxis2
border[_range0:_range1,0] = xmin
border[_range0:_range1,1] = yside
#right
_range0 = _range1
_range1 = _range0 + naxis2
border[_range0:_range1,0] = xmax
border[_range0:_range1,1] = yside
edges = wcs.all_pix2world(border[:,0],border[:,1],1)
return edges
def make_radio_combination_signature(radio_annotation, wcs, atlas_positions,
subject, pix_offset):
"""Generates a unique signature for a radio annotation.
radio_annotation: 'radio' dictionary from a classification.
wcs: World coordinate system associated with the ATLAS image.
atlas_positions: [[RA, DEC]] NumPy array.
subject: RGZ subject dict.
pix_offset: (x, y) pixel position of this radio subject on the ATLAS image.
-> Something immutable
"""
from . import rgz_data as data
# TODO(MatthewJA): This only works on ATLAS. Generalise.
# My choice of immutable object will be stringified crowdastro ATLAS
# indices.
zooniverse_id = subject['zooniverse_id']
subject_fits = data.get_radio_fits(subject)
subject_wcs = astropy.wcs.WCS(subject_fits.header)
atlas_ids = []
x_offset, y_offset = pix_offset
for c in radio_annotation.values():
# Note that the x scale is not the same as the IR scale, but the scale
# factor is included in the annotation, so I have multiplied this out
# here for consistency.
scale_width = c.get('scale_width', '')
scale_height = c.get('scale_height', '')
if scale_width:
scale_width = float(scale_width)
else:
# Sometimes, there's no scale, so I've included a default scale.
scale_width = config['surveys']['atlas']['scale_width']
if scale_height:
scale_height = float(scale_height)
else:
scale_height = config['surveys']['atlas']['scale_height']
# These numbers are in terms of the PNG images, so I need to multiply by
# the click-to-fits ratio.
scale_width *= config['surveys']['atlas']['click_to_fits_x']
scale_height *= config['surveys']['atlas']['click_to_fits_y']
subject_bbox = [
[
float(c['xmin']) * scale_width,
float(c['xmax']) * scale_width,
],
[
float(c['ymin']) * scale_height,
float(c['ymax']) * scale_height,
],
]
# ...and by the mosaic ratio. There's probably double-up here, but this
# makes more sense.
scale_width *= config['surveys']['atlas']['mosaic_scale_x']
scale_height *= config['surveys']['atlas']['mosaic_scale_y']
# Get the bounding box of the radio source in pixels.
# Format: [xs, ys]
bbox = [
[
float(c['xmin']) * scale_width,
float(c['xmax']) * scale_width,
],
[
float(c['ymin']) * scale_height,
float(c['ymax']) * scale_height,
],
]
assert bbox[0][0] < bbox[0][1]
assert bbox[1][0] < bbox[1][1]
# Convert the bounding box into RA/DEC.
bbox = wcs.all_pix2world(bbox[0] + x_offset, bbox[1] + y_offset,
FITS_CONVENTION)
subject_bbox = subject_wcs.all_pix2world(subject_bbox[0],
subject_bbox[1], FITS_CONVENTION)
# TODO(MatthewJA): Remove (or disable) this sanity check.
# The bbox is backwards along the x-axis for some reason.
bbox[0] = bbox[0][::-1]
assert bbox[0][0] < bbox[0][1]
assert bbox[1][0] < bbox[1][1]
bbox = numpy.array(bbox)
# What is this radio source called? Check if we have an object in the
# bounding box. We'll cache these results because there is a lot of
# overlap.
cache_key = tuple(tuple(b) for b in bbox)
if cache_key in bbox_cache_:
index = bbox_cache_[cache_key]
else:
x_gt_min = atlas_positions[:, 0] >= bbox[0, 0]
x_lt_max = atlas_positions[:, 0] <= bbox[0, 1]
y_gt_min = atlas_positions[:, 1] >= bbox[1, 0]
y_lt_max = atlas_positions[:, 1] <= bbox[1, 1]
within = numpy.all([x_gt_min, x_lt_max, y_gt_min, y_lt_max], axis=0)
indices = numpy.where(within)[0]
if len(indices) == 0:
logging.debug('Skipping radio source not in catalogue for '
'%s', zooniverse_id)
continue
else:
if len(indices) > 1:
logging.debug('Found multiple (%d) ATLAS matches '
'for %s', len(indices), zooniverse_id)
index = indices[0]
bbox_cache_[cache_key] = index
atlas_ids.append(str(index))
atlas_ids.sort()
if not atlas_ids:
raise CatalogueError('No catalogued radio sources.')
return ';'.join(atlas_ids)
(check_sources, check_dead, check_sky,
check_source_sky) = check_hdulists
pyfits.HDUList(check_sources).writeto("check_sources.fits",
overwrite=True)
pyfits.HDUList(check_dead).writeto("check_dead.fits", overwrite=True)
pyfits.HDUList(check_sky).writeto("check_sky.fits", overwrite=True)
pyfits.HDUList(check_source_sky).writeto("check_source_sky.fits",
overwrite=True)
# src_data.info()
# convert polygon center coordinates from native pixels to Ra/Dec
#src_data.info()
#print(src_data['center_x'].astype(numpy.float).to_numpy())
_ra, _dec = wcs.all_pix2world(
src_data['center_x'].astype(numpy.float).to_numpy(),
src_data['center_y'].astype(numpy.float).to_numpy(), 1)
src_data['center_ra'] = _ra
src_data['center_dec'] = _dec
#print(radec)
# apply flux calibrations
calib_factor = 1.0
if (name in calibration_factors):
calib_factor = calibration_factors[name]
named_logger.info("Apply calibration factor: %g" % (calib_factor))
sky_error = (src_data['src_area'] * src_data['sky_var']).astype(
numpy.float).to_numpy()
src_error = gain * src_data['src_flux'].astype(numpy.float).to_numpy()
flx_error = numpy.fabs(src_error) + sky_error * gain**2
def detect_sources(image, cat_name=None):
""" A generator of (ra, dec) tuples (PROOF OF CONCEPT) """
log.debug("Reading FITS file")
with astropy.io.fits.open(image) as hdulist:
data = hdulist[0].data
header = hdulist[0].header
with warnings.catch_warnings():
warnings.simplefilter("ignore")
log.info("Loading WCS data")
wcs = astropy.wcs.WCS(header)
try:
log.info("Estimating background")
background = sextractor.Background(data)
except ValueError:
# Fix for error "Input array with dtype '>f4' has non-native
# byte order. Only native byte order arrays are supported".
log.debug("Converting data to native byte order")
data = data.byteswap(True).newbyteorder()
background = sextractor.Background(data)
log.info("Subtracting background")
background.subfrom(data) # in-place
# Global "average" RMS of background
log.info("Detecting sources")
rms_ntimes = 1.5
while True:
try:
threshold = rms_ntimes * background.globalrms
objects = sextractor.extract(data, threshold)
except Exception as e:
# Fix for error "internal pixel buffer full: The limit of 300000
# active object pixels over the detection threshold was reached.
# Check that the image is background subtracted and the detection
# threshold is not too low. detection threshold". If a different
# exception is raised, just re-raise it.
if "threshold is not too low" in str(e):
rms_ntimes *= 1.25
log.debug("Internal pixel buffer full")
log.debug("Retrying with threshold {0:.2} times RMS of background".format(rms_ntimes))
else:
raise
else:
break
pixel_coords = numpy.empty([0,2])
print "len(objects) = ", len(objects)
print "Image = ", image
sys.exit()
# https://github.com/kbarbary/sep/blob/master/sep.pyx#L555
log.info("Transforming pixel to celestial coordinates")
for index in range(len(objects)):
x, y = objects[index]['x'], objects[index]['y']
pixel_coords = numpy.row_stack(pixel_coords, (x, y))
print pixel_coords.shape
# If cat_name is present, save the coordinates to a file
if cat_name:
fd = open(cat_name, 'w')
for ra, dec in list(wcs.all_pix2world(pixel_coords[:,0], pixel_coords[:,1], 0)):
fd.write( "{0} {1} \n ".format(ra, dec) )
fd.close()
return list(wcs.all_pix2world(pixel_coords[:,0], pixel_coords[:,1], 0))
basedir,
uncompressed=args.uncompressed)
else:
print('No bright star catalog, not marking bright stars.')
res = process(im,
sqivar,
flag,
psf,
refit_psf=args.refit_psf,
verbose=args.verbose,
nx=4,
ny=4,
derivcentroids=True)
outfn = args.outfn[0]
x = (res[0])['x']
y = (res[0])['y']
wcs = wcs.WCS(hdr)
ra, dec = wcs.all_pix2world(y, x, 0)
import numpy.lib.recfunctions as rfn
cat = rfn.append_fields(res[0], ['ra', 'dec'], [ra, dec])
fits.writeto(outfn, cat)
fits.append(outfn, res[1])
fits.append(outfn, res[2])
psfext = numpy.array([tpsf(0, 0, stampsz=19) for tpsf in res[3]])
fits.append(outfn, psfext)
def make_radio_combination_signature(radio_annotation, wcs, atlas_positions,
subject, pix_offset):
"""Generates a unique signature for a radio annotation.
radio_annotation: 'radio' dictionary from a classification.
wcs: World coordinate system associated with the ATLAS image.
atlas_positions: [[RA, DEC]] NumPy array.
subject: RGZ subject dict.
pix_offset: (x, y) pixel position of this radio subject on the ATLAS image.
-> Something immutable
"""
from . import rgz_data as data
# TODO(MatthewJA): This only works on ATLAS. Generalise.
# My choice of immutable object will be stringified crowdastro ATLAS
# indices.
zooniverse_id = subject['zooniverse_id']
subject_fits = data.get_radio_fits(subject)
subject_wcs = astropy.wcs.WCS(subject_fits.header)
atlas_ids = []
x_offset, y_offset = pix_offset
for c in radio_annotation.values():
# Note that the x scale is not the same as the IR scale, but the scale
# factor is included in the annotation, so I have multiplied this out
# here for consistency.
scale_width = c.get('scale_width', '')
scale_height = c.get('scale_height', '')
if scale_width:
scale_width = float(scale_width)
else:
# Sometimes, there's no scale, so I've included a default scale.
scale_width = config['surveys']['atlas']['scale_width']
if scale_height:
scale_height = float(scale_height)
else:
scale_height = config['surveys']['atlas']['scale_height']
# These numbers are in terms of the PNG images, so I need to multiply by
# the click-to-fits ratio.
scale_width *= config['surveys']['atlas']['click_to_fits_x']
scale_height *= config['surveys']['atlas']['click_to_fits_y']
subject_bbox = [
[
float(c['xmin']) * scale_width,
float(c['xmax']) * scale_width,
],
[
float(c['ymin']) * scale_height,
float(c['ymax']) * scale_height,
],
]
# ...and by the mosaic ratio. There's probably double-up here, but this
# makes more sense.
scale_width *= config['surveys']['atlas']['mosaic_scale_x']
scale_height *= config['surveys']['atlas']['mosaic_scale_y']
# Get the bounding box of the radio source in pixels.
# Format: [xs, ys]
bbox = [
[
float(c['xmin']) * scale_width,
float(c['xmax']) * scale_width,
],
[
float(c['ymin']) * scale_height,
float(c['ymax']) * scale_height,
],
]
assert bbox[0][0] < bbox[0][1]
assert bbox[1][0] < bbox[1][1]
# Convert the bounding box into RA/DEC.
bbox = wcs.all_pix2world(bbox[0] + x_offset, bbox[1] + y_offset,
FITS_CONVENTION)
subject_bbox = subject_wcs.all_pix2world(subject_bbox[0],
subject_bbox[1],
FITS_CONVENTION)
# TODO(MatthewJA): Remove (or disable) this sanity check.
# The bbox is backwards along the x-axis for some reason.
bbox[0] = bbox[0][::-1]
assert bbox[0][0] < bbox[0][1]
assert bbox[1][0] < bbox[1][1]
bbox = numpy.array(bbox)
# What is this radio source called? Check if we have an object in the
# bounding box. We'll cache these results because there is a lot of
# overlap.
cache_key = tuple(tuple(b) for b in bbox)
if cache_key in bbox_cache_:
index = bbox_cache_[cache_key]
else:
x_gt_min = atlas_positions[:, 0] >= bbox[0, 0]
x_lt_max = atlas_positions[:, 0] <= bbox[0, 1]
y_gt_min = atlas_positions[:, 1] >= bbox[1, 0]
y_lt_max = atlas_positions[:, 1] <= bbox[1, 1]
within = numpy.all([x_gt_min, x_lt_max, y_gt_min, y_lt_max],
axis=0)
indices = numpy.where(within)[0]
if len(indices) == 0:
logging.debug(
'Skipping radio source not in catalogue for '
'%s', zooniverse_id)
continue
else:
if len(indices) > 1:
logging.debug(
'Found multiple (%d) ATLAS matches '
'for %s', len(indices), zooniverse_id)
index = indices[0]
bbox_cache_[cache_key] = index
atlas_ids.append(str(index))
atlas_ids.sort()
if not atlas_ids:
raise CatalogueError('No catalogued radio sources.')
return ';'.join(atlas_ids)
(_median - 3 * _sigma))
use_for_fit[bad] = False
area_ellipse = numpy.pi * a * b
# break
fit_results.append([a, b, theta, x0, y0, area, area_ellipse])
df = pandas.DataFrame(data=fit_results,
columns=[
'A', 'B', 'theta', 'x0', 'y0',
'area_contours', "area_ellipse"
])
wcs = astropy.wcs.WCS(header)
radec = wcs.all_pix2world(df[['x0', 'y0']].to_numpy(), 0)
print(radec)
df.loc[:, ['ra', 'dec']] = radec
df.loc[:, 'filename'] = fn
df.loc[:, ['a_arcsec', 'b_arcsec']] = df[['A', 'B'
]].to_numpy() * pixelscale
df.loc[:, 'area_error'] = (df['area_ellipse'] -
df['area_contours']) / df['area_contours']
df.loc[:, 'good_fit'] = numpy.fabs(df['area_error']) < 0.05
df.info()
if (master_df is None):
master_df = df
else:
master_df = master_df.append(df, ignore_index=False)
# now write the region file again, this time using WCS instead of pixel information
hdr = img_hdu[0].header
try:
pixelscale = numpy.hypot(hdr['CD1_1'],
hdr['CD1_2']) * 3600. # in arcsec/pixel
except:
try:
pixelscale = hdr['CDELT1'] * 3600
except:
pixelscale = 1.
pass
# convert all x/y into ra/dec
wcs = astropy.wcs.WCS(hdr)
(ra0, dec0) = wcs.all_pix2world(df['x0'], df['y0'], 0)
# print(ra_dec)
df['ra0'] = ra0 #_dec[:,0]
df['dec0'] = dec0 #ra_dec[:,1]
ds9_reg_fn = output_basename + "_ellipses_wcs.reg"
with open(ds9_reg_fn, "w") as ds9_reg:
print("""\
# Region file format: DS9 version 4.1
global color=green dashlist=8 3 width=1 font="helvetica 10 normal roman" select=1 highlite=1 dash=0 fixed=0 edit=1 move=1 delete=1 include=1 source=1
fk5""",
file=ds9_reg)
for index, e in df.iterrows():
# print(e)
print('ellipse(%f,%f,%f",%f",%f) # %s' %
(e['ra0'], e['dec0'], e['sma'] * pixelscale,
(1. - e['ellipticity']) * e['sma'] * pixelscale, e['pa'],