def match_2dhist_fit(imglist, reference_catalog, **fit_pars):
"""Perform cross-matching and final fit using 2dHistogram matching
Parameters
----------
imglist : list
List of input image `~tweakwcs.tpwcs.FITSWCS` objects with metadata and source catalogs
reference_catalog : Table
Astropy Table of reference sources for this field
Returns
--------
imglist : list
List of input image `~tweakwcs.tpwcs.FITSWCS` objects with metadata and source catalogs
"""
log.info("{} (match_2dhist_fit) Cross matching and fitting "
"{}".format("-" * 20, "-" * 28))
# Specify matching algorithm to use
match = tweakwcs.TPMatch(**fit_pars)
# Align images and correct WCS
tweakwcs.align_wcs(imglist,
reference_catalog,
match=match,
expand_refcat=False)
# Interpret RMS values from tweakwcs
interpret_fit_rms(imglist, reference_catalog)
return imglist
def match_relative_fit(imglist, reference_catalog, **fit_pars):
"""Perform cross-matching and final fit using relative matching algorithm
Parameters
----------
imglist : list
List of input image `~tweakwcs.tpwcs.FITSWCS` objects with metadata and source catalogs
reference_catalog : Table
Astropy Table of reference sources for this field
Returns
--------
imglist : list
List of input image `~tweakwcs.tpwcs.FITSWCS` objects with metadata and source catalogs
"""
log.info("{} (match_relative_fit) Cross matching and fitting {}".format(
"-" * 20, "-" * 27))
if 'fitgeom' in fit_pars:
fitgeom = fit_pars['fitgeom']
del fit_pars['fitgeom']
else:
fitgeom = 'general'
# 0: Specify matching algorithm to use
match = tweakwcs.TPMatch(**fit_pars)
# match = tweakwcs.TPMatch(searchrad=250, separation=0.1,
# tolerance=100, use2dhist=False)
# Align images and correct WCS
# NOTE: this invocation does not use an astrometric catalog. This call allows all the input images to be aligned in
# a relative way using the first input image as the reference.
# 1: Perform relative alignment
tweakwcs.align_wcs(imglist,
None,
match=match,
expand_refcat=True,
fitgeom=fitgeom)
# Set all the group_id values to be the same so the various images/chips will be aligned to the astrometric
# reference catalog as an ensemble.
# astrometric reference catalog as an ensemble. BEWARE: If additional iterations of solutions are to be
# done, the group_id values need to be restored.
for image in imglist:
image.meta["group_id"] = 1234567
# 2: Perform absolute alignment
tweakwcs.align_wcs(imglist,
reference_catalog,
match=match,
fitgeom=fitgeom)
# 3: Interpret RMS values from tweakwcs
interpret_fit_rms(imglist, reference_catalog)
return imglist
def match_default_fit(imglist, reference_catalog, **fit_pars):
"""Perform cross-matching and final fit using default tolerance matching
Parameters
----------
imglist : list
List of input image `~tweakwcs.tpwcs.FITSWCS` objects with metadata and source catalogs
reference_catalog : Table
Astropy Table of reference sources for this field
Returns
--------
imglist : list
List of input image `~tweakwcs.tpwcs.FITSWCS` objects with metadata and source catalogs
"""
if 'fitgeom' in fit_pars:
fitgeom = fit_pars['fitgeom']
del fit_pars['fitgeom']
else:
fitgeom = 'rscale'
common_pars = fit_pars['pars']
del fit_pars['pars']
log.info("{} (match_default_fit) Cross matching and fitting "
"{}".format("-" * 20, "-" * 27))
# Specify matching algorithm to use
match = tweakwcs.TPMatch(**fit_pars)
# Align images and correct WCS
matched_cat = tweakwcs.align_wcs(imglist,
reference_catalog,
match=match,
minobj=common_pars['MIN_FIT_MATCHES'],
expand_refcat=False,
fitgeom=fitgeom)
# Interpret RMS values from tweakwcs
interpret_fit_rms(imglist, reference_catalog)
del matched_cat
return imglist
def match_relative_fit(imglist, reference_catalog, **fit_pars):
"""Perform cross-matching and final fit using relative matching algorithm
Parameters
----------
imglist : list
List of input image `~tweakwcs.tpwcs.FITSWCS` objects with metadata and source catalogs
reference_catalog : Table
Astropy Table of reference sources for this field
Returns
--------
imglist : list
List of input image `~tweakwcs.tpwcs.FITSWCS` objects with metadata and source catalogs
"""
log.info("{} (match_relative_fit) Cross matching and fitting {}".format(
"-" * 20, "-" * 27))
if 'fitgeom' in fit_pars:
fitgeom = fit_pars['fitgeom']
del fit_pars['fitgeom']
else:
fitgeom = 'rscale'
common_pars = fit_pars['pars']
del fit_pars['pars']
# 0: Specify matching algorithm to use
match = tweakwcs.TPMatch(**fit_pars)
# match = tweakwcs.TPMatch(searchrad=250, separation=0.1,
# tolerance=100, use2dhist=False)
# Align images and correct WCS
# NOTE: this invocation does not use an astrometric catalog. This call
# allows all the input images to be aligned in
# a relative way using the first input image as the reference.
# 1: Perform relative alignment
match_relcat = tweakwcs.align_wcs(imglist,
None,
match=match,
minobj=common_pars['MIN_FIT_MATCHES'],
expand_refcat=True,
fitgeom=fitgeom)
log.info("Relative alignment found: ")
for i in imglist:
info = i.meta['fit_info']
if 'shift' not in info:
off = (0., 0.)
rot = 0.
scale = -1.
nmatches = 0
else:
off = info['shift']
rot = info['<rot>']
scale = info['<scale>']
nmatches = info['nmatches']
msg = "Image {} --".format(i.meta['name'])
msg += "\n SHIFT:({:9.4f},{:9.4f}) NMATCHES: {} ".format(
off[0], off[1], nmatches)
msg += "\n ROT:{:9.4f} SCALE:{:9.4f}".format(rot, scale)
msg += "\n Using fitgeom = '{}'".format(fitgeom)
log.info(msg)
# This logic enables performing only relative fitting and skipping fitting to GAIA
if reference_catalog is not None:
# Set all the group_id values to be the same so the various images/chips will be aligned to the astrometric
# reference catalog as an ensemble.
# astrometric reference catalog as an ensemble. BEWARE: If additional iterations of solutions are to be
# done, the group_id values need to be restored.
for image in imglist:
image.meta["group_id"] = 1234567
# 2: Perform absolute alignment
matched_cat = tweakwcs.align_wcs(imglist,
reference_catalog,
match=match,
minobj=common_pars['MIN_FIT_MATCHES'],
fitgeom=fitgeom)
else:
# Insure the expanded reference catalog has all the information needed
# to complete processing.
# TODO: Work out how to get the 'mag' column from input source catalog
# into this extended reference catalog...
reference_catalog = match_relcat
reference_catalog['mag'] = np.array([-999.9] * len(reference_catalog),
np.float32)
reference_catalog.meta['catalog'] = 'relative'
# 3: Interpret RMS values from tweakwcs
interpret_fit_rms(imglist, reference_catalog)
del match_relcat
return imglist
def determine_alignment_residuals(input,
files,
catalogs=None,
max_srcs=1000,
json_timestamp=None,
json_time_since_epoch=None,
log_level=logutil.logging.INFO):
"""Determine the relative alignment between members of an association.
Parameters
-----------
input : string
Original pipeline input filename. This filename will be used to
define the output analysis results filename.
files : list
Set of files on which to actually perform comparison. The original
pipeline can work on both CTE-corrected and non-CTE-corrected files,
but this comparison will only be performed on CTE-corrected
products when available.
catalogs : list, optional
List of dictionaries containing the source catalogs for each input chip.
The list NEEDS to be in the same order as the filenames given in `files`.
Each dictionary for each file will need to have numerical (integer) keys
for each 'sci' extension. If left as `None`, this function will create
it's own set of catalogs using `astrometric_utils.extract_point_sources`.
json_timestamp: str, optional
Universal .json file generation date and time (local timezone) that will be used in the instantiation
of the HapDiagnostic object. Format: MM/DD/YYYYTHH:MM:SS (Example: 05/04/2020T13:46:35). If not
specified, default value is logical 'None'
json_time_since_epoch : float
Universal .json file generation time that will be used in the instantiation of the HapDiagnostic
object. Format: Time (in seconds) elapsed since January 1, 1970, 00:00:00 (UTC). If not specified,
default value is logical 'None'
log_level : int, optional
The desired level of verboseness in the log statements displayed on the screen and written to the
.log file. Default value is 'NOTSET'.
Returns
--------
resids_files : list of string
Name of JSON files containing all the extracted results from the comparisons
being performed.
"""
log.setLevel(log_level)
if catalogs is None:
# Open all files as HDUList objects
hdus = [fits.open(f) for f in files]
# Determine sources from each chip
src_cats = []
num_srcs = []
for hdu in hdus:
numsci = countExtn(hdu)
nums = 0
img_cats = {}
if hdu[("SCI", 1)].data.max() == 0.0:
log.info(
"SKIPPING point-source finding for blank image: {}".format(
hdu.filename()))
continue
log.info("Determining point-sources for {}".format(hdu.filename()))
for chip in range(numsci):
chip += 1
img_cats[chip] = amutils.extract_point_sources(
hdu[("SCI", chip)].data, nbright=max_srcs)
nums += len(img_cats[chip])
log.info("Identified {} point-sources from {}".format(
nums, hdu.filename()))
num_srcs.append(nums)
src_cats.append(img_cats)
else:
src_cats = catalogs
num_srcs = []
for img in src_cats:
num_img = 0
for chip in img:
num_img += len(img[chip])
num_srcs.append(num_img)
if len(num_srcs) == 0 or (len(num_srcs) > 0 and max(num_srcs) <= 3):
log.warning(
"Not enough sources identified in input images for comparison")
return []
# Combine WCS from HDULists and source catalogs into tweakwcs-compatible input
imglist = []
for i, (f, cat) in enumerate(zip(files, src_cats)):
imglist += amutils.build_wcscat(f, i, cat)
# Setup matching algorithm using parameters tuned to well-aligned images
match = tweakwcs.TPMatch(searchrad=5,
separation=4.0,
tolerance=1.0,
use2dhist=True)
try:
# perform relative fitting
matchlist = tweakwcs.align_wcs(imglist,
None,
minobj=6,
match=match,
expand_refcat=False)
del matchlist
except Exception:
try:
# Try without 2dHist use to see whether we can get any matches at all
match = tweakwcs.TPMatch(searchrad=5,
separation=4.0,
tolerance=1.0,
use2dhist=False)
matchlist = tweakwcs.align_wcs(imglist,
None,
minobj=6,
match=match,
expand_refcat=False)
del matchlist
except Exception:
log.warning("Problem encountered during matching of sources")
return []
# Check to see whether there were any successful fits...
align_success = False
for img in imglist:
wcsname = fits.getval(img.meta['filename'], 'wcsname', ext=("sci", 1))
img.meta['wcsname'] = wcsname
img.meta['fit_info']['aligned_to'] = imglist[0].meta['filename']
img.meta['reference_catalog'] = None
for img in imglist:
if img.meta['fit_info']['status'] == 'SUCCESS' and '-FIT' in wcsname:
align_success = True
break
resids_files = []
if align_success:
# extract results in the style of 'tweakreg'
resids = extract_residuals(imglist)
if resids is not None:
resids_files = generate_output_files(
resids,
json_timestamp=json_timestamp,
json_time_since_epoch=json_time_since_epoch,
exclude_fields=['group_id'])
return resids_files
def test_multichip_fitswcs_alignment():
h1 = fits.Header.fromfile(get_pkg_data_filename('data/wfc3_uvis1.hdr'))
w1 = wcs.WCS(h1)
imcat1 = tweakwcs.FITSWCS(w1)
imcat1.meta['catalog'] = table.Table.read(
get_pkg_data_filename('data/wfc3_uvis1.cat'),
format='ascii.csv',
delimiter=' ',
names=['x', 'y'])
imcat1.meta['group_id'] = 1
imcat1.meta['name'] = 'ext1'
h2 = fits.Header.fromfile(get_pkg_data_filename('data/wfc3_uvis2.hdr'))
w2 = wcs.WCS(h2)
imcat2 = tweakwcs.FITSWCS(w2)
imcat2.meta['catalog'] = table.Table.read(
get_pkg_data_filename('data/wfc3_uvis2.cat'),
format='ascii.csv',
delimiter=' ',
names=['x', 'y'])
imcat2.meta['group_id'] = 1
imcat2.meta['name'] = 'ext4'
refcat = table.Table.read(get_pkg_data_filename('data/ref.cat'),
format='ascii.csv',
delimiter=' ',
names=['RA', 'DEC'])
tweakwcs.align_wcs([imcat1, imcat2],
refcat,
match=_match,
nclip=None,
sigma=3,
fitgeom='general')
fi1 = imcat1.meta['fit_info']
fi2 = imcat2.meta['fit_info']
w1 = imcat1.wcs
w2 = imcat2.wcs
assert np.allclose(w1.wcs.crval, (83.206917667519, -67.73275818507248),
rtol=0)
assert np.allclose(
w1.wcs.cd,
np.array([[3.93222694902149e-06, -1.0106698270131359e-05],
[-1.0377001075437075e-05, -4.577945148472431e-06]]),
atol=0.0,
rtol=1e-8)
assert np.allclose(w2.wcs.crval, (83.15167050722597, -67.74220306069903),
rtol=0)
assert np.allclose(
w2.wcs.cd,
np.array([[3.834449806681195e-06, -9.996495217498745e-06],
[-1.0348147451241423e-05, -4.503496019301529e-06]]),
atol=0.0,
rtol=1e-8)
assert np.allclose(fi1['<scale>'], 1.0025, rtol=0, atol=2e-8)
assert np.allclose(fi2['<scale>'], 1.0025, rtol=0, atol=2e-8)
assert fi1['rmse'] < 5e-5
assert fi2['rmse'] < 5e-5
cat1 = imcat1.meta['catalog']
ra1, dec1 = w1.all_pix2world(cat1['x'], cat1['y'], 0)
cat2 = imcat2.meta['catalog']
ra2, dec2 = w2.all_pix2world(cat2['x'], cat2['y'], 0)
ra = np.concatenate([ra1, ra2])
dec = np.concatenate([dec1, dec2])
rmse_ra = np.sqrt(np.mean((ra - refcat['RA'])**2))
rmse_dec = np.sqrt(np.mean((dec - refcat['DEC'])**2))
assert rmse_ra < 1e-9
assert rmse_dec < 1e-9
def test_different_ref_tpwcs_fitswcs_alignment(wcsno, refscale, dra, ddec):
# This test was designed to check that the results of alignment,
# in particular and most importantly, the sky positions of sources in
# aligned images do not depend on the tangent reference plane used
# for alignment. [#125]
h1 = fits.Header.fromfile(get_pkg_data_filename('data/wfc3_uvis1.hdr'))
w1 = wcs.WCS(h1)
imcat1 = tweakwcs.FITSWCS(w1)
imcat1.meta['catalog'] = table.Table.read(
get_pkg_data_filename('data/wfc3_uvis1.cat'),
format='ascii.csv',
delimiter=' ',
names=['x', 'y'])
imcat1.meta['group_id'] = 1
imcat1.meta['name'] = 'ext1'
h2 = fits.Header.fromfile(get_pkg_data_filename('data/wfc3_uvis2.hdr'))
w2 = wcs.WCS(h2)
imcat2 = tweakwcs.FITSWCS(w2)
imcat2.meta['catalog'] = table.Table.read(
get_pkg_data_filename('data/wfc3_uvis2.cat'),
format='ascii.csv',
delimiter=' ',
names=['x', 'y'])
imcat2.meta['group_id'] = 1
imcat2.meta['name'] = 'ext4'
refcat = table.Table.read(get_pkg_data_filename('data/ref.cat'),
format='ascii.csv',
delimiter=' ',
names=['RA', 'DEC'])
refwcses = [wcs.WCS(h1), wcs.WCS(h2)]
refwcs = refwcses[wcsno]
# change pointing of the reference WCS (alignment tangent plane):
refwcs.wcs.crval = refwcses[1 - wcsno].wcs.crval + np.asarray([dra, ddec])
rotm = tweakwcs.linearfit.build_fit_matrix(*refscale)
refwcs.wcs.cd = np.dot(refwcs.wcs.cd, rotm)
refwcs.wcs.set()
ref_tpwcs = tweakwcs.FITSWCS(refwcs)
tweakwcs.align_wcs([imcat1, imcat2],
refcat,
ref_tpwcs=ref_tpwcs,
match=_match,
nclip=None,
sigma=3,
fitgeom='general')
fi1 = imcat1.meta['fit_info']
fi2 = imcat2.meta['fit_info']
w1 = imcat1.wcs
w2 = imcat2.wcs
assert fi1['rmse'] < 1e-4
assert fi2['rmse'] < 1e-4
cat1 = imcat1.meta['catalog']
ra1, dec1 = w1.all_pix2world(cat1['x'], cat1['y'], 0)
cat2 = imcat2.meta['catalog']
ra2, dec2 = w2.all_pix2world(cat2['x'], cat2['y'], 0)
ra = np.concatenate([ra1, ra2])
dec = np.concatenate([dec1, dec2])
rmse_ra = np.sqrt(np.mean((ra - refcat['RA'])**2))
rmse_dec = np.sqrt(np.mean((dec - refcat['DEC'])**2))
assert rmse_ra < 5e-9
assert rmse_dec < 5e-9
def test_multichip_jwst_alignment():
w1 = _make_gwcs_wcs('data/wfc3_uvis1.hdr')
imcat1 = tweakwcs.JWSTgWCS(w1, {'v2_ref': 0, 'v3_ref': 0, 'roll_ref': 0})
imcat1.meta['catalog'] = table.Table.read(
get_pkg_data_filename('data/wfc3_uvis1.cat'),
format='ascii.csv',
delimiter=' ',
names=['x', 'y'])
imcat1.meta['catalog']['x'] += 1
imcat1.meta['catalog']['y'] += 1
imcat1.meta['group_id'] = 1
imcat1.meta['name'] = 'ext1'
w2 = _make_gwcs_wcs('data/wfc3_uvis2.hdr')
imcat2 = tweakwcs.JWSTgWCS(w2, {'v2_ref': 0, 'v3_ref': 0, 'roll_ref': 0})
imcat2.meta['catalog'] = table.Table.read(
get_pkg_data_filename('data/wfc3_uvis2.cat'),
format='ascii.csv',
delimiter=' ',
names=['x', 'y'])
imcat2.meta['catalog']['x'] += 1
imcat2.meta['catalog']['y'] += 1
imcat2.meta['group_id'] = 1
imcat2.meta['name'] = 'ext4'
refcat = table.Table.read(get_pkg_data_filename('data/ref.cat'),
format='ascii.csv',
delimiter=' ',
names=['RA', 'DEC'])
tweakwcs.align_wcs([imcat1, imcat2],
refcat,
match=_match,
nclip=None,
sigma=3,
fitgeom='general')
fi1 = imcat1.meta['fit_info']
fi2 = imcat2.meta['fit_info']
w1m = imcat1.wcs
w2m = imcat2.wcs
assert np.allclose(w1m(*w1.crpix), (83.206917667519, -67.73275818507248),
rtol=0)
assert np.allclose(w2m(*w2.crpix), (83.15167050722597, -67.74220306069903),
rtol=0)
assert np.allclose(fi1['<scale>'], 1.0025, rtol=0, atol=2e-8)
assert np.allclose(fi2['<scale>'], 1.0025, rtol=0, atol=2e-8)
assert fi1['rmse'] < 5e-5
assert fi2['rmse'] < 5e-5
ra1, dec1 = imcat1.wcs(imcat1.meta['catalog']['x'],
imcat1.meta['catalog']['y'])
ra2, dec2 = imcat2.wcs(imcat2.meta['catalog']['x'],
imcat2.meta['catalog']['y'])
ra = np.concatenate([ra1, ra2])
dec = np.concatenate([dec1, dec2])
rra = refcat['RA']
rdec = refcat['DEC']
rmse_ra = np.sqrt(np.mean((ra - rra)**2))
rmse_dec = np.sqrt(np.mean((dec - rdec)**2))
assert rmse_ra < 3e-9
assert rmse_dec < 3e-10
def determine_alignment_residuals(input, files, max_srcs=2000):
"""Determine the relative alignment between members of an association.
Parameters
-----------
input : string
Original pipeline input filename. This filename will be used to
define the output analysis results filename.
files : list
Set of files on which to actually perform comparison. The original
pipeline can work on both CTE-corrected and non-CTE-corrected files,
but this comparison will only be performed on CTE-corrected
products when available.
Returns
--------
results : string
Name of JSON file containing all the extracted results from the comparisons
being performed.
"""
# Open all files as HDUList objects
hdus = [fits.open(f) for f in files]
# Determine sources from each chip
src_cats = []
num_srcs = []
for hdu in hdus:
numsci = countExtn(hdu)
nums = 0
img_cats = {}
for chip in range(numsci):
chip += 1
img_cats[chip] = amutils.extract_point_sources(hdu[("SCI",
chip)].data,
nbright=max_srcs)
nums += len(img_cats[chip])
num_srcs.append(nums)
src_cats.append(img_cats)
if len(num_srcs) == 0 or (len(num_srcs) > 0 and max(num_srcs) <= 3):
return None
# src_cats = [amutils.generate_source_catalog(hdu) for hdu in hdus]
# Combine WCS from HDULists and source catalogs into tweakwcs-compatible input
imglist = []
for i, (f, cat) in enumerate(zip(files, src_cats)):
imglist += amutils.build_wcscat(f, i, cat)
# Setup matching algorithm using parameters tuned to well-aligned images
match = tweakwcs.TPMatch(searchrad=5,
separation=1.0,
tolerance=4.0,
use2dhist=True)
try:
# perform relative fitting
matchlist = tweakwcs.align_wcs(imglist,
None,
match=match,
expand_refcat=False)
del matchlist
except Exception:
return None
# Check to see whether there were any successful fits...
align_success = False
for img in imglist:
if img.meta['fit_info']['status'] == 'SUCCESS':
align_success = True
break
if align_success:
# extract results in the style of 'tweakreg'
resids = extract_residuals(imglist)
if resids is None:
resids_file = None
else:
# Define name for output JSON file...
resids_file = "{}_astrometry_resids.json".format(input[:9])
# Remove any previously computed results
if os.path.exists(resids_file):
os.remove(resids_file)
# Dump the results to a JSON file now...
with open(resids_file, 'w') as jfile:
json.dump(resids, jfile)
else:
resids_file = None
return resids_file