def measure_backgrounds(cat_table, ref_im):
"""
Measure the background for all the sources in cat_table, using
ref_im.
Parameters
----------
cat_table: astropy Table
the catalog in astropy table form, as loaded from a .gst.fits
photometry catalog
ref_im: imageHDU
fits image which will be used to estimate the background
Returns
-------
measurements: list of float
a metric for the background intensity at the star's position
(photometry / area)
mask: 2d ndarray of bool
an array containing a bool for each pixel of the image, True if
the pixel was ignored for the background calculations
"""
if not ref_im:
# Return a dumb value
return cat_table['F814W_CHI']
w = wcs.WCS(ref_im.header)
shp = ref_im.data.shape
inner_rad = 30 * units.pixel
outer_rad = inner_rad + 20 * units.pixel
mask_rad = inner_rad
# More elaborate way using an actual image (do not care about the
# filter for the moment, just take the image (which was given on the
# command line) at face value)
ra = cat_table['RA']
dec = cat_table['DEC']
c = astropy.coordinates.SkyCoord(ra * units.degree, dec * units.degree)
# Annuli, of which the counts per surface area will be used as
# background measurements
annuli = pu.SkyCircularAnnulus(c, r_in=inner_rad, r_out=outer_rad)
area = annuli.to_pixel(w).area()
# A mask to make sure that no sources end up in the background
# calculation
circles = pu.SkyCircularAperture(c, mask_rad)
source_masks = circles.to_pixel(w).to_mask()
mask_union = np.zeros(shp)
for i, ap_mask in enumerate(source_masks):
data_slices = list(ap_mask.bbox.slices)
# These slices go outside of the box sometimes! In that case we
# will override the slices on both sides.
# Default slices (go over the whole mask)
mask_slices = [slice(None, None), slice(None, None)]
# Adjust the slices for x and y if necessary
for j in range(2):
# DATA: . . a b c d e f
# index - - 0 1 2 3 4 5
# ---------------
# MASK: - + + + -
# index 0 1 2 3 4
# --> DATA_SLICE [0:stop]
# --> MASK_SLICE [2:]
data_start = data_slices[j].start
if data_start < 0:
# Move the start from negative n to zero
data_slices[j] = slice(0, data_slices[j].stop)
# Move the start from 0 to positive n
mask_slices[j] = slice(-data_start, mask_slices[j].stop)
# --> we slice over the part of the small mask that
# falls on the positive side of the axis
data_stop = data_slices[j].stop
if data_stop > shp[j]:
overflow = data_stop - shp[j]
# Move the stop 'overflow' to shp[j]
data_slices[j] = slice(data_slices[j].start, shp[j])
# Move the stop to 'overflow' pixels from the end
mask_slices[j] = slice(mask_slices[j].start, -overflow)
# --> slice over the part that falls below the maximum
mask_union[data_slices] += ap_mask.data[mask_slices]
# Threshold
mask_union = mask_union > 0
phot = pu.aperture_photometry(ref_im.data, annuli, wcs=w, mask=mask_union)
return phot['aperture_sum'] / area, mask_union
def measure_backgrounds(cat_table, ref_im, mask_radius, ann_width, cat_filter):
"""
Measure the background for all the sources in cat_table, using
ref_im.
Parameters
----------
cat_table: astropy Table
the catalog in astropy table form, as loaded from a .gst.fits
photometry catalog
ref_im: imageHDU
fits image which will be used to estimate the background
mask_radius : float
radius (in pixels) of mask for catalog sources
ann_width : float
width of annulus (in pixels) for calculating background around each catalog source
cat_filter : list or None
If list: Two elements in which the first is a filter (e.g. 'F475W') and
the second is a magnitude. Catalog entries with [filter]_VEGA > mag
will not be masked.
If None: all catalog entries will be considered.
Returns
-------
measurements: list of float
a metric for the background intensity at the star's position
(photometry / area)
"""
w = wcs.WCS(ref_im.header)
shp = ref_im.data.shape
inner_rad = mask_radius * units.pixel
outer_rad = inner_rad + ann_width * units.pixel
mask_rad = inner_rad
# More elaborate way using an actual image (do not care about the
# filter for the moment, just take the image (which was given on the
# command line) at face value)
ra = cat_table['RA']
dec = cat_table['DEC']
c = astropy.coordinates.SkyCoord(ra * units.degree, dec * units.degree)
# Annuli, of which the counts per surface area will be used as
# background measurements
annuli = pu.SkyCircularAnnulus(c, r_in=inner_rad, r_out=outer_rad)
area = annuli.to_pixel(w).area()
# A mask to make sure that no sources end up in the background
# calculation
if cat_filter is None:
circles = pu.SkyCircularAperture(c, mask_rad)
else:
circles = pu.SkyCircularAperture(c[ cat_table[cat_filter[0]+'_VEGA'] < float(cat_filter[1]) ], mask_rad)
source_masks = circles.to_pixel(w).to_mask()
mask_union = np.zeros(shp)
for i, ap_mask in enumerate(source_masks):
# the masks have a bounding box which we need to take into
# account. Here we will calculate the overlap between the mask
# and the image (important if the mask is near the edge, so that
# the box might go outside of it).
data_slices = list(ap_mask.bbox.slices)
# These slices go outside of the box sometimes! In that case we
# will override the slices on both sides.
# Default slices (go over the whole mask)
mask_slices = [slice(None, None), slice(None, None)]
# Adjust the slices in each dimension
for j in range(2):
# DATA: . . a b c d e f
# index - - 0 1 2 3 4 5
# ---------------
# MASK: - + + + -
# index 0 1 2 3 4
# --> DATA_SLICE [0:stop]
# --> MASK_SLICE [2:]
data_start = data_slices[j].start
if data_start < 0:
# Move the start from negative n to zero
data_slices[j] = slice(0, data_slices[j].stop)
# Move the start from 0 to positive n
mask_slices[j] = slice(-data_start, mask_slices[j].stop)
# --> we slice over the part of the small mask that
# falls on the positive side of the axis
data_stop = data_slices[j].stop
if data_stop > shp[j]:
overflow = data_stop - shp[j]
# Move the stop 'overflow' to shp[j]
data_slices[j] = slice(data_slices[j].start, shp[j])
# Move the stop to 'overflow' pixels from the end
mask_slices[j] = slice(mask_slices[j].start, -overflow)
# --> slice over the part that falls below the maximum
mask_union[tuple(data_slices)] += ap_mask.data[tuple(mask_slices)]
# Threshold
mask_union = mask_union > 0
# also mask NaNs
mask_union[np.isnan(ref_im.data)] = True
# Save the masked reference image
hdu = fits.PrimaryHDU(np.where(mask_union, 0, ref_im.data), header=ref_im.header)
hdu.writeto('masked_reference_image.fits', overwrite=True)
# Do the measurements
phot = pu.aperture_photometry(ref_im.data, annuli, wcs=w, mask=mask_union)
return phot['aperture_sum'] / area