def cutFits(f, ra, dec, size=(4 * u.arcsec, 4 * u.arcsec), sigma=False):
if not os.path.isfile(f): return None
try:
if len(size) != 2:
return
except Exception as e:
print(
"\033[31msize must be an int or length-2 list/tuple/array\033[0m"
)
frame = fFile[0].header['SYSTEM'].strip()
p = SkyCoord(ra=ra * u.degree, dec=dec * u.degree, frame=frame.lower())
sizeQuant = u.Quantity(size, u.arcsec)
wcs = WCS(header=fFile[0].header)
try:
cutout = Cutout2D(fFile[0].data, p, sizeQuant, wcs=wcs)
if sigma:
sigma = getSigma(fFile)
sigma_cutout = Cutout2D(sigma, p, size, wcs=wcs)
return cutout.data, sigma_cutout.data
else:
return cutout.data
except NoOverlapError:
print(
'Ra, Dec not inside frame for Ra:' +
' {ra}, Dec: {dec}, and frame: {f}'.format(
ra=ra, dec=dec, f=fFile[0]
)
)
return False
def _get_local_cutout(self):
"""Fetch cutout data via local FITS images (e.g. RACS / VLASS)."""
fields = self._find_image()
assert len(
fields
) > 0, f"No fields located at {self.position.ra:.2f}, {self.position.dec:.2f}"
closest = fields[fields.dist_field_centre ==
fields.dist_field_centre.min()].iloc[0]
image_path = SURVEYS.loc[self.survey]['images']
if self.survey == 'vlass':
filepath = f'{closest.epoch}/{closest.tile}/{closest.image}/{closest.filename}'
image_path = vlass_path
elif 'racs' in self.survey:
pol = self.survey[-1]
if on_system == 'ada':
filepath = f'RACS_test4_1.05_{closest.field}.fits'
else:
filepath = f'RACS_{closest.field}.EPOCH00.{pol}.fits'
elif 'vast' in self.survey:
pattern = re.compile(r'vastp(\dx*)([IV])')
epoch = pattern.sub(r'\1', self.survey)
pol = pattern.sub(r'\2', self.survey)
filepath = f'VAST_{closest.field}.EPOCH0{epoch}.{pol}.fits'
else:
filepath = f'*{closest.field}*0.restored.fits'
try:
self.filepath = glob.glob(image_path + filepath)[0]
except IndexError:
raise FITSException(
f'Could not match {self.survey} image filepath: \n{image_path + filepath}'
)
with fits.open(self.filepath) as hdul:
self.header, data = hdul[0].header, hdul[0].data
wcs = WCS(self.header, naxis=2)
self.mjd = Time(self.header['DATE']).mjd
try:
cutout = Cutout2D(data[0, 0, :, :],
self.position,
self.radius * u.deg,
wcs=wcs)
except IndexError:
cutout = Cutout2D(data,
self.position,
self.radius * u.deg,
wcs=wcs)
self.data = cutout.data * 1000
self.wcs = cutout.wcs
if 'racs' in self.survey or 'vast' in self.survey:
self.pos_err = SURVEYS.loc[self.basesurvey].pos_err
self._get_source()
else:
# Probably using vlass, yet to include aegean catalogs
self.plot_sources = False
self.plot_neighbours = False
def plot_residual_image(old_res, new_res, mode):
nres = fits.getdata(new_res)
ores = fits.getdata(old_res)
cmap = 'hot'
cnorm = SymLogNorm(linthresh=30, linscale=0.5)
#
# Zoomed out
#
fig = plt.figure(figsize=(10, 5))
ax1 = plt.subplot(121, aspect='equal')
im = plt.imshow(ores, cmap=cmap, norm=cnorm)
plt.title('Old ' + mode)
plt.gca().invert_yaxis()
plt.subplot(122, sharex=ax1, sharey=ax1, aspect='equal')
plt.imshow(nres, cmap=cmap, norm=cnorm)
plt.title('New ' + mode)
plt.gca().invert_yaxis()
plt.subplots_adjust(right=0.8)
cbar_ax = fig.add_axes([0.85, 0.15, 0.05, 0.7])
fig.colorbar(im, cax=cbar_ax)
#
# Zoomed in.
#
c_pos = [500, 550]
c_siz = [200, 200]
nres_c = Cutout2D(nres, c_pos, c_siz)
ores_c = Cutout2D(ores, c_pos, c_siz)
cnorm = SymLogNorm(linthresh=30,
linscale=0.5,
vmin=nres_c.data.min(),
vmax=nres_c.data.max())
fig2 = plt.figure(figsize=(10, 5))
ax2 = plt.subplot(121, aspect='equal')
im = plt.imshow(ores_c.data,
extent=np.array(ores_c.bbox_original)[::-1].flatten(),
cmap=cmap,
norm=cnorm)
plt.title('Old ' + mode)
plt.gca().invert_yaxis()
plt.subplot(122, sharex=ax2, sharey=ax2, aspect='equal')
plt.imshow(nres_c.data,
extent=np.array(nres_c.bbox_original)[::-1].flatten(),
cmap=cmap,
norm=cnorm)
plt.title('New ' + mode)
plt.gca().invert_yaxis()
plt.subplots_adjust(right=0.8)
cbar_ax = fig2.add_axes([0.85, 0.15, 0.05, 0.7])
fig2.colorbar(im, cax=cbar_ax)
plt.show()
return
def detect(self):
cs = [SkyCoord(ra,dec,unit='deg') for ra,dec in zip(self.ras,self.decs)]
cw = WCS(self.cheader)
if self.wv > 240:
cw_n = WCS(self.nheader)
size = u.Quantity((self.pixsize,self.pixsize),u.degree)
fluxes = [Cutout2D(self.cmap,c,size,wcs = cw).data for c in cs]
else:
size =
if self.wv > 240:
noises = [Cutout2D(self.noise,c,size,wcs = cw_n).data for c in cs]
else:
noises = [self.noise for i in range(len(cs))]
dd_fluxes = []
dd_noises = []
dd_ras = []
dd_decs = []
dd_mags = []
for i in range(len(fluxes)):
if fluxes[i] > 2*noises[i]:
dd_fluxes.append(fluxes[i])
dd_noises.append(noises[i])
dd_ras.append(self.ras[i])
dd_decs.append(self.decs[i])
dd_mags.append(self.mags[i])
results = Table([dd_ras,dd_decs,dd_fluxes,dd_noises,dd_mags],names = ['RA','DEC','flux','noise','PetroMag_i'])
return results
def Load_image(self):
if self.file is None:
raise ValueError('No file specified')
try:
hdu = fits.open(self.file)
except:
raise ValueError('Could not load {}'.format(self.file))
self.date = self.file.split('tess')[-1].split('-')[0]
self.sector = self.file.split('-')[1]
self.camera = self.file.split('-')[2]
self.ccd = self.file.split('-')[3]
self.hdu = hdu[1]
self.header = hdu[1].header
data = hdu[1].data
err = hdu[2].data
self.wcs = WCS(hdu[1].header)
cut = Cutout2D(data, (1024 + 44, 1024), 2048, wcs=self.wcs)
self.image = cut.data
wcs = cut.wcs
err = Cutout2D(err, (1024 + 44, 1024), 2048).data
self.noise = err
print('Successfully loaded {}'.format(self.file))
return
def get_next_im(im_locs, obj_loc, min_visible_region, cutout_dims):
for i in im_locs:
print(i)
im = fits.open(i)[-1]
wcs = WCS(im.header)
pix_xy = wcs.all_world2pix(obj_loc, 1)[0] #Pixel coords of obj
if ((min_visible_region != 0)&(min_visible_region !=(0,0))): #minimal visible region wanted
try: #Should only succeed if all pixels in this cutout have values
dummy = Cutout2D(im.data, pix_xy, min_visible_region, mode='strict')
except:
#print('Image at obsjd {} rejected due to NaN close to im center'.format(im.header['OBSJD']))
continue #Start next i in loop
cutout = Cutout2D(im.data, pix_xy, cutout_dims, mode='partial', wcs=wcs)
#Get relevand image data not available in the cutout
im_center = [round(pix_xy[0]), round(pix_xy[1])]
if 'ZTFData/ref' in i:
imtype = 'ref'
date = ' '
if im.header['DBFID']==1:
obsfilter = 'ZTF_g'
elif im.header['DBFID']==2:
obsfilter = 'ZTF_r'
else:
obsfilter = 'ZTF_i'
else:
imtype = 'dif'
date = time.Time(im.header['OBSJD'], format='jd')
obsfilter = im.header['FILTER']
#yield 1 item which is a list of required items
yield [cutout, imtype, date, obsfilter, im_center]
def make_cutout(image,
WCS_img,
position,
size=100.0,
size_in_arcsec=False,
show_img=False):
'''
Makes cutout.
'''
# get coordinate frame
WCS_img.sip = None
# define position
if size_in_arcsec:
size = u.Quantity((size, size), u.arcsec)
cutout = Cutout2D(image, position, size, wcs=WCS_img)
else:
size = (size * u.pixel, size * u.pixel)
cutout = Cutout2D(
image,
WCS_img.wcs_world2pix([[position.ra.deg, position.dec.deg]], 1)[0],
size)
if show_img:
img = pl.imshow(cutout.data, origin='lower')
pl.axis('off')
pl.show()
return (cutout)
def get_next_im(target):
'''
Generator function for making the animation
The next image in the list is opened, the target location is found and cut
out, and the cutout is tested. If the tests are passed successfully, the
cutout will be used in the next frame of the animation. Else a warning is
issued before continueing to the next image.
1st test: Is the minimum visible region fully observed (e.g. no NaN)?
2nd test: Are there positive values? (prevents weird negative frames found
when using previous versions of the code)
Parameters:
target (anim_target): Object to animate.
Yields:
list: A list containing the rotated cutout, rotation details, and
some metadata.
'''
for i in target.im_locs:
#Open next image and check if it meets the animation requirements.
im = fits.open(i)[-1]
wcs = WCS(im.header)
pix_xy = wcs.all_world2pix([(target.ra, target.dec)], 0)[0]
#1st test
if (('/sci/' in i) & (target.min_visible!=0) & (target.min_visible!=(0,0))):
try: #Should only succeed if all pixels in this cutout have values
dummy = Cutout2D(im.data, pix_xy, target.min_visible,
mode='strict')
except: #Can't fully fill min visible region
print('Image at obsjd {} rejected due to NaN'\
' close to im center'.format(im.header['OBSJD']))
continue #Start next i in loop
cutout = Cutout2D(im.data, pix_xy, target.cutout_dims, mode='partial',
wcs=wcs)
#2nd test
if np.nanmax(cutout.data) < 0:
print('Image at obsjd {} rejected due to all values being negative\
'.format(im.header['OBSJD']))
continue
#Get needed info from the header (Give entire header in the future?)
if 'ZTFData/ref' in i:
imtype = 'ref'
else:
imtype = 'dif'
#Correct image orientation / being mirrored if needed.
immat, rots, flip, resid = find_orientation(cutout)
im_center = [round(pix_xy[0]), round(pix_xy[1])]
#yield a list of items required by the subplots
yield [immat, imtype, im.header, rots, flip, resid, im_center]
def getTailPeaks(tailTimeSteps, noradid, mwa, sat, ts, args):
"""
finds all the peak pixels in tail using gaussian fit
Parameters
----------
tailTimeSteps : the list of tail time steps
noradid : the norad id of the satellite
mwa : observer object
sat : the satellite object
ts : time object
Returns
-------
x_array : array of x pixels of tail
y_array : array of y pixels of tail
ra_array : array of ra of tail
dec_array : array of dec of tail
"""
x_array, y_array = [], []
x_err_array, y_err_array = [], []
y = np.linspace(0, 4, 5)
x = np.linspace(0, 4, 5)
x, y = np.meshgrid(x, y)
for t in tailTimeSteps:
hdu = fits.open("Neg" + args.filePrefix + "-t" + str(t).zfill(4) + ".fits")
UTCTime = datetime.strptime(hdu[0].header['DATE-OBS'], '%Y-%m-%dT%H:%M:%S.%f')
data = hdu[0].data
data = maskData(data, UTCTime, mwa, sat, wcs, ts)
if np.all(data == 0):
print("file full of zeros")
print("aborting....")
sys.exit(0)
row, col = np.where(data == np.min(data))
beam_cutout = Cutout2D(beam, (col, row), (5 ,5 ))
cutout = Cutout2D(data, (col, row), (5,5))
temp = cutout.data / beam_cutout.data
temp /= np.nansum(temp)
initial_guess = (temp[1,1], 1, 1, 2)
popt, pconv = opt.curve_fit(twoD_Gaussian, (x, y), temp.ravel(), p0=initial_guess)
perr = np.sqrt(np.diag(pconv))
mu_x, mu_y = popt[1], popt[2]
sigma_x, sigma_y = perr[1], perr[2]
solX = float(col + (mu_x - 2))
solY = float(row + (mu_y - 2))
x_array.append(solX)
y_array.append(solY)
x_err_array.append(sigma_x)
y_err_array.append(sigma_y)
return x_array, y_array, x_err_array, y_err_array
def low_bkg_ref(self):
files = self.imagefiles
summed = np.zeros(len(files)) * np.nan
ex = self.hduext
for i in range(len(files)):
hdu = fits.open(files[i])
data = hdu[ex].data
wcs = WCS(hdu[ex].header)
cut = Cutout2D(data, (1024 + 44, 1024), 2048, wcs=wcs)
data = cut.data
wcs = cut.wcs
data[data <= 0] = np.nan
if np.nansum(abs(data)) > 0:
summed[i] = np.nansum(abs(data))
lim = np.percentile(summed[np.isfinite(summed)], 5)
ind = np.where((summed < lim))[0]
good = files[ind]
goods = np.zeros((len(good), 2048, 2048))
var = np.zeros((len(good), 2048, 2048))
mjd = np.zeros(len(good))
i = 0
sat_count = np.zeros_like(data)
for g in good:
hdu = fits.open(g)
data = hdu[ex].data
wcs = WCS(hdu[ex].header)
cut = Cutout2D(data, (1024 + 44, 1024), 2048, wcs=wcs)
data = cut.data
wcs = cut.wcs
goods[i] = data
e = hdu[2].data
cut = Cutout2D(e, (1024 + 44, 1024), 2048)
data = cut.data
var[i] = data**2
jd = hdu[1].header['TSTART'] + hdu[1].header['BJDREFI']
mjd[i] = Time(jd, format='jd', scale='tdb').mjd
sat_count[data > 4.8E4 - 500] += 1
i += 1
ref = np.nanmedian(goods, axis=0)
var = np.nanmedian(var, axis=0)
hdu[1].header['MJD'] = (np.nanmean(mjd), 'stacked')
hdu[1].header['NIMAGES'] = (str(len(good)), 'number of images stacked')
sats = sat_count / len(good) >= (0.05)
self.hdu = hdu[1]
self.header = hdu[1].header
self.image = ref
self.noise = np.sqrt(var)
return
def image_snippet(image,
center,
width=50,
axis=None,
fig=None,
is_mask=False,
pad_black=False,
**kwargs):
"""
Display a subsection of an image about a center.
Parameters
----------
image : numpy array
The full image from which a section is to be taken.
center : list-like
The location of the center of the cutout.
width : int, optional
Width of the cutout, in pixels.
axis : matplotlib.Axes instance, optional
Axis on which the image should be displayed.
fig : matplotlib.Figure, optional
Figure on which the image should be displayed.
is_mask : bool, optional
Set to ``True`` if the image is a mask, i.e. all values are
either zero or one.
pad_black : bool, optional
If ``True``, pad edges of the image with zeros to fill out width
if the slice is near the edge.
"""
if pad_black:
sub_image = Cutout2D(image,
center,
width,
mode='partial',
fill_value=0)
else:
# Return a smaller subimage if extent goes out side image
sub_image = Cutout2D(image, center, width, mode='trim')
show_image(sub_image.data,
cmap='gray',
ax=axis,
fig=fig,
show_colorbar=False,
show_ticks=False,
is_mask=is_mask,
**kwargs)
def Reference_image(files,FITS=False): summed = np.zeros(len(files)) * np.nan for i in range(len(files)): hdu = fits.open(files[i]) data = hdu[1].data wcs = WCS(hdu[1].header) cut = Cutout2D(data,(1024+44,1024),2048,wcs=wcs) data = cut.data wcs = cut.wcs data[data <= 0] = np.nan if np.nansum(abs(data)) > 0: summed[i] = np.nansum(abs(data)) lim = np.percentile(summed[np.isfinite(summed)],5) ind = np.where((summed < lim))[0] good = files[ind] goods = np.zeros((len(good),2048,2048)) var = np.zeros((len(good),2048,2048)) mjd = np.zeros(len(good)) i = 0 for g in good: hdu = fits.open(g) data = hdu[1].data wcs = WCS(hdu[1].header) cut = Cutout2D(data,(1024+44,1024),2048,wcs=wcs) data = cut.data wcs = cut.wcs goods[i] = data e = hdu[2].data cut = Cutout2D(e,(1024+44,1024),2048) data = cut.data var[i] = data**2 jd = hdu[1].header['TSTART'] + hdu[1].header['BJDREFI'] mjd[i] = Time(jd, format='jd', scale='tdb').mjd i += 1 ref = np.nanmedian(goods,axis=0) var = np.nanmedian(var,axis=0) hdu[1].header['MJD'] = (np.nanmean(mjd), 'stacked') if FITS: ref_fits = deepcopy(hdu) ref_fits[1].data = ref ref_fits[1].header.update(wcs.to_header()) ref_fits[2].data = err ref_fits[2].header.update(wcs.to_header()) return ref_fits else: return ref, np.sqrt(var), wcs, hdu
def get_cutout(path, name, position, size):
sci = fits.open(os.path.join(path, name+"sci.fits"))
wht = fits.open(os.path.join(path, name+"wht.fits"))
image = sci[0].data
image = image.byteswap().newbyteorder()
weight = wht[0].data
weight = weight.byteswap().newbyteorder()
wcs = WCS(sci[0].header)
cutout_image = Cutout2D(image, position, size, wcs=wcs)
cutout_weight = Cutout2D(weight, position, size, wcs=wcs)
image = np.ascontiguousarray(cutout_image.data)
weight = np.ascontiguousarray(cutout_weight.data)
rms = np.sqrt(1.0 / weight)
return image, weight, rms, cutout_image, cutout_image.wcs
def test_copy(self):
data = np.copy(self.data)
c = Cutout2D(data, (2, 3), (3, 3))
xy = (0, 0)
value = 100.
c.data[xy] = value
xy_orig = c.to_original_position(xy)
yx = xy_orig[::-1]
assert data[yx] == value
data = np.copy(self.data)
c2 = Cutout2D(self.data, (2, 3), (3, 3), copy=True)
c2.data[xy] = value
assert data[yx] != value
def getTailPeaks(tailTimeSteps, noradid, mwa, sat, ts):
x_array, y_array, ra_array, dec_array = [], [], [], []
hduBeam = fits.open(beamFile)
beam = hduBeam[0].data
y = np.linspace(0, 4, 5)
x = np.linspace(0, 4, 5)
x, y = np.meshgrid(x, y)
for t in tailTimeSteps:
hdu = fits.open("Neg" +
"6Sigma1Floodfilllr14SigmaRFIBinaryMapPeakFlux-t" +
str(t).zfill(4) + ".fits")
UTCTime = datetime.strptime(hdu[0].header['DATE-OBS'],
'%Y-%m-%dT%H:%M:%S.%f')
data = hdu[0].data
data = maskData(data, UTCTime, mwa, sat, wcs, ts)
if np.all(data == 0):
tailTimeSteps.remove(t)
continue
row, col = np.where(data == np.min(data))
beam_cutout = Cutout2D(beam, (col, row), (5, 5))
cutout = Cutout2D(data, (col, row), (5, 5))
temp = cutout.data / beam_cutout.data
temp /= abs(np.sum(temp))
initial_guess = (temp[1, 1], 1, 1, 2)
popt, pconv = opt.curve_fit(twoD_Gaussian, (x, y),
temp.ravel(),
p0=initial_guess)
perr = np.sqrt(np.diag(pconv))
mu_x, mu_y = popt[1], popt[2]
print("tail t {} x {} y {} dx {} dy {}".format(t, mu_x, mu_y, perr[1],
perr[2]))
solX = col + (mu_x - 2)
solY = row + (mu_y - 2)
x_array.append(solX[0])
y_array.append(solY[0])
pixcrd = np.array([[0, 0], [solX, solY]], dtype=np.float64)
world = wcs.wcs_pix2world(pixcrd, 0)
ra, dec = world[1]
ra_array.append(ra)
dec_array.append(dec)
return x_array, y_array, ra_array, dec_array
def gen_cutouts(cutout_dims, radec, refquery, difquery, obj_name):
nr_refims = len(refquery.metatable)
nr_frames = nr_refims + len(
difquery.metatable) #amount of images to go through
#Make an empty list containing cutouts
cutouts = []
#Also define an array storing relevant data for each image
im_dat = pd.DataFrame(
columns=[
'obj_name', 'im_type', 'obs_filter', 'obsjd', 'peak_val',
'mean_val', 'standev'
],
index=np.arange(0, nr_frames)) #Probably still needs to be expanded
#Loop through the images to get required data
data_locs = refquery.get_local_data() + difquery.get_local_data(
'scimrefdiffimg.fits.fz')
for i in range(len(data_locs)):
im = fits.open(data_locs[i])
wcs = WCS(im[-1].header)
pix_xy = wcs.all_world2pix(radec, 1)[0] #Pixel location of object
try: #See if the target coord is in the image, if not go to next image
dummy = Cutout2D(im[-1].data, pix_xy, 1, mode='strict')
except:
continue
cutouts.append(
Cutout2D(im[-1].data, pix_xy, cutout_dims,
mode='partial')) #Do the cutout, pad if needed
#record data about image
j = np.shape(cutouts)[0] - 1 #Current index is that of the last cutout
im_dat.obj_name[j] = obj_name
if i < nr_refims:
im_dat.im_type[j] = 'ref'
im_dat.obsjd[j] = ' '
if refquery.metatable.fid[j] == 1:
im_dat.obs_filter[j] = 'ZTF_g'
elif refquery.metatable.fid[j] == 2:
im_dat.obs_filter[j] = 'ZTF_r'
else:
im_dat.obs_filter[j] = 'ZTF_i'
else:
im_dat.im_type[j] = 'diff'
im_dat.obsjd[j] = im[-1].header['OBSJD']
im_dat.obs_filter[j] = im[-1].header['FILTER']
im_dat.peak_val[j] = np.nanmax(cutouts[j].data)
im_dat.mean_val[j] = np.nanmean(cutouts[j].data)
im_dat.standev[j] = np.nanstd(cutouts[j].data)
#Close fits file before moving on to the next
im.close()
return cutouts, im_dat
def isolated_center(x, y, image):
"""Finds the centroid of each isolated source with centroid_quadratic.
Parameters
----------
x : array-like
Initial guesses of x positions for sources.
y : array-like
Initial guesses of y positions for sources.
image : np.ndarray
FFI flux data.
Returns
-------
cenx : array-like
Controid x coordinates for all good input sources.
ceny : array-like
Centroid y coordinates for all good input sources.
good : list
Indexes into input arrays corresponding to good sources.
"""
cenx, ceny, good = [], [], []
for i in range(len(x)):
if x[i] > 0. and y[i] > 0.:
tpf = Cutout2D(image, position=(x[i], y[i]), size=(7,7), mode='partial')
origin = tpf.origin_original
cen = centroid_quadratic(tpf.data - np.nanmedian(tpf.data))
cenx.append(cen[0]+origin[0]); ceny.append(cen[1]+origin[1])
good.append(i)
cenx, ceny = np.array(cenx), np.array(ceny)
return cenx, ceny, good
def get_postage_stamp(radecstr, size, image, image_wcs):
"""
Extract a postage stamp image from a larger image
Parameters
----------
radecstr : str
SkyCoord pareseable string for ra, dec coordinates
size : 2 element tuple
size in ra and dec in arcsec
image : 2D numpy.ndarray
image to cut postage stamps from
image_wcs : astropy.wcs
WCS for image
"""
position = SkyCoord(radecstr)
pix_scales = 3600. * proj_plane_pixel_scales(image_wcs)
pix_size = (size[0] / pix_scales[0], size[1] / pix_scales[1])
try:
result = Cutout2D(image, position, pix_size, wcs=image_wcs)
except:
return None
return result
def cutout(self, position, width, mode="trim"):
"""
Create a cutout around a given position.
Parameters
----------
position : `~astropy.coordinates.SkyCoord`
Center position of the cutout region.
width : tuple of `~astropy.coordinates.Angle`
Angular sizes of the region in (lon, lat) in that specific order.
If only one value is passed, a square region is extracted.
mode : {'trim', 'partial', 'strict'}
Mode option for Cutout2D, for details see `~astropy.nddata.utils.Cutout2D`.
Returns
-------
cutout : `~gammapy.maps.WcsNDMap`
Cutout map
"""
width = _check_width(width)
dummy_data = np.empty(self.to_image().data_shape)
c2d = Cutout2D(
data=dummy_data,
wcs=self.wcs,
position=position,
# Cutout2D takes size with order (lat, lon)
size=width[::-1] * u.deg,
mode=mode,
)
return self._init_copy(wcs=c2d.wcs, npix=c2d.shape[::-1])
def _get_stamps(fits_file, ra, dec, size, indiv_fits_files=None, return_indiv_stamps=False):
"""
Cut out science, template, and difference stamps around a detection at `ra,dec` from `fits_file`
"""
def get_indiv_stamps():
if not indiv_fits_files:
return
_indiv_stamps = np.zeros((size, size, len(indiv_fits_files)))
for j, indiv_fits_file in enumerate(indiv_fits_files):
try:
im_wcs = WCS(indiv_fits_file[SCIENCE_IMAGE_EXT].header)
stamp_coo = SkyCoord(ra, dec, unit='deg')
indiv_cutout = Cutout2D(indiv_fits_file[SCIENCE_IMAGE_EXT].data, stamp_coo, (size, size),
mode='partial', fill_value=1e-6, wcs=im_wcs).data
except KeyError:
logger.warning("extension {} missing from {}".format(image_ext, fits_file))
return
except NoOverlapError:
logger.warning("requested stamp(s) are off edge of image")
return
_indiv_stamps[:, :, j] = indiv_cutout
return _indiv_stamps
# layers are: science, template, difference, peak-to-peak individual, min individual, max individual
stamps = np.zeros((size, size, 6))
for i, image_ext in enumerate([SCIENCE_IMAGE_EXT, TEMPLATE_IMAGE_EXT, DIFFERENCE_IMAGE_EXT]):
try:
im_wcs = WCS(fits_file[image_ext].header)
stamp_coo = SkyCoord(ra, dec, unit='deg')
cutout = Cutout2D(fits_file[image_ext].data, stamp_coo, (size, size),
mode='partial', fill_value=1e-6, wcs=im_wcs, ).data
except KeyError:
logger.warning("extension {} missing from {}".format(image_ext, fits_file))
return
except NoOverlapError:
logger.warning("requested stamp is off edge of image")
return
stamps[:, :, i] = cutout
indiv_stamps = get_indiv_stamps()
# modify return signature to support multiarg.
if (indiv_stamps is None) and ~return_indiv_stamps:
return stamps
if (indiv_stamps is None) and return_indiv_stamps:
return stamps, None
# Add additional frames that capture statistics from the individual science images
stamps[:, :, 3] = np.ptp(np.atleast_3d(indiv_stamps), axis=2)
stamps[:, :, 4] = np.min(np.atleast_3d(indiv_stamps), axis=2)
stamps[:, :, 5] = np.max(np.atleast_3d(indiv_stamps), axis=2)
if return_indiv_stamps:
return stamps, indiv_stamps
else:
return stamps
def cut_2d(data_masked,position,size,wcs):
data=data_masked.data
cut=Cutout2D(data=data,position=position,size=size,wcs=wcs)
data_cut=cut.data
wcs_cut=cut.wcs
return wcs_cut, data_cut
def centroiding_iteration(ccd, position_xy, cbox_size=5., csigma=3.):
''' Find the intensity-weighted centroid of the image iteratively
Returns
-------
xc_img, yc_img : float
The centroided location in the original image coordinate in image XY.
shift : float
The total distance between the initial guess and the fitted centroid,
i.e., the distance between `(xc_img, yc_img)` and `position_xy`.
'''
imgX, imgY = position_xy
cutccd = Cutout2D(ccd.data, position=position_xy, size=cbox_size)
avg, med, std = sigma_clipped_stats(cutccd.data, sigma=3, iters=5)
cthresh = med + csigma * std
# using pixels only above med + 3*std for centroiding is recommended.
# See Ma+2009, Optics Express, 17, 8525
mask = (cutccd.data < cthresh)
if ccd.mask is not None:
mask += ccd.mask
xc_cut, yc_cut = centroid_com(data=cutccd.data, mask=mask)
# Find the centroid with pixels have values > 3sigma, by center of mass
# method. The position is in the cutout image coordinate, e.g., (3, 3).
xc_img, yc_img = cutccd.to_original_position((xc_cut, yc_cut))
# convert the cutout image coordinate to original coordinate.
# e.g., (3, 3) becomes something like (137, 189)
dx = xc_img - imgX
dy = yc_img - imgY
shift = np.sqrt(dx**2 + dy**2)
return xc_img, yc_img, shift
def cutout2CCDData(ccd,
position,
size,
mode='trim',
fill_value=np.nan,
full=True):
''' Converts the Cutout2D object to proper CCDData.
Parameters
----------
ccd: CCDData
The ccd to be trimmed.
position, size, mode, fill_value:
See the ``ccdproc.trim_image`` doc.
full: bool
If ``True`` (default), returns the ``Cutout2D`` object.
If ``False``, only the trimmed ccd is returned.
'''
w = WCS(ccd.header)
cut = Cutout2D(data=ccd.data,
position=position,
size=size,
wcs=w,
mode=mode,
fill_value=fill_value,
copy=True)
# Copy True just to avoid any contamination to the original ccd.
y1 = cut.ymin_original
y2 = cut.ymax_original
x1 = cut.xmin_original
x2 = cut.xmax_original
trimmed_ccd = trim_image(ccd[y1:y2, x1:x2])
if full:
return trimmed_ccd, cut
return trimmed_ccd
def cut_hdu(self, location, size, writetodisk=False, saveas=None):
"""
cutout size lxw ~ (dy x dx) box on fits file centered at a pixel or skycoord location
if size is scalar gives a square dy=dx
updates hdr wcs keeps other info from original
"""
try:
hdu = self.sci
except:
hdu = self
cphdu = hdu.copy()
dat = cphdu.data
hdr = cphdu.header
wcs, frame = WCS(hdr), hdr['RADESYS'].lower()
print("cut_hdu")
print("location,wcs", location, wcs)
print("shape", dat.shape)
cut = Cutout2D(dat, location, size, wcs=wcs)
cutwcs = cut.wcs
cphdu.data = cut.data
cphdu.header.update(cut.wcs.to_header())
if writetodisk:
cphdu.writeto(saveas, overwrite=True)
self.postage_stamp = cphdu
return cphdu
def single_cutout(ax,position,image,mask1=None,mask2=None,points=None,label=None,size=6*u.arcsec):
cutout_image = Cutout2D(image.data,position,size=size,wcs=image.wcs)
norm = simple_norm(cutout_image.data,clip=False,stretch='linear',percent=99.5)
ax.imshow(cutout_image.data,origin='lower',norm=norm,cmap=plt.cm.gray_r)
# plot the nebulae catalogue
cutout_mask, _ = reproject_interp(mask1,output_projection=cutout_image.wcs,shape_out=cutout_image.shape,order='nearest-neighbor')
region_ID = np.unique(cutout_mask[~np.isnan(cutout_mask)])
contours = []
for i in region_ID:
blank_mask = np.zeros_like(cutout_mask)
blank_mask[cutout_mask==i] = 1
contours += find_contours(blank_mask, 0.5)
for coords in contours:
ax.plot(coords[:,1],coords[:,0],color='tab:red',lw=1,label='HII-region')
mask = np.zeros((*cutout_mask.shape,4))
mask[~np.isnan(cutout_mask.data),:] = (0.84, 0.15, 0.16,0.1)
ax.imshow(mask,origin='lower')
# plot the association catalogue
if mask2:
cutout_mask, _ = reproject_interp(mask2,output_projection=cutout_image.wcs,shape_out=cutout_image.shape,order='nearest-neighbor')
region_ID = np.unique(cutout_mask[~np.isnan(cutout_mask)])
contours = []
for i in region_ID:
blank_mask = np.zeros_like(cutout_mask)
blank_mask[cutout_mask==i] = 1
contours += find_contours(blank_mask, 0.5)
for coords in contours:
ax.plot(coords[:,1],coords[:,0],color='tab:blue',lw=1,label='association')
mask = np.zeros((*cutout_mask.shape,4))
mask[~np.isnan(cutout_mask.data),:] = (0.12,0.47,0.71,0.1)
ax.imshow(mask,origin='lower')
# mark the position of the clusters within the cutout
if points:
region = RectangleSkyRegion(position,0.9*size,0.9*size)
in_frame = points[region.contains(points['SkyCoord'],cutout_image.wcs)]
for row in in_frame:
x,y = row['SkyCoord'].to_pixel(cutout_image.wcs)
if 5<x<cutout_image.data.shape[0]-5 and 5<y<cutout_image.data.shape[1]-5:
ax.scatter(x,y,marker='o',facecolors='none',s=20,lw=0.4,color='tab:blue',label='cluster')
if label:
t = ax.text(0.07,0.875,label, transform=ax.transAxes,color='black',fontsize=8)
t.set_bbox(dict(facecolor='white', alpha=1, ec='white'))
ax.set_xticks([])
ax.set_yticks([])
return ax
def apply_recenter_cutout(self, ra, dec):
"""
Center the image at ra, dec. The image will be cut accordingly.
Parameters
----------
ra: float, RA in deg.
dec: float, Dec in deg.
"""
new_center_pix = self.get_wcs().all_world2pix([[ra, dec]], 0)[0]
new_center_skycoord = SkyCoord(ra, dec, unit="deg")
shape = np.array(np.shape(self.img_data))
new_shape = np.array([
2 * np.min([shape[0] - new_center_pix[0], new_center_pix[0]]),
2 * np.min([shape[1] - new_center_pix[1], new_center_pix[1]])
]).astype(int)
print('ncp', new_center_pix, new_center_pix[::-1], ra, dec)
print(self.img_hdr['CRVAL1'], self.img_hdr['CRVAL2'])
self.img_hdr['CRVAL1']
cutout = Cutout2D(self.img_data.T,
new_center_pix.T,
new_shape.T,
wcs=self.get_wcs().copy(),
copy=True) #, mode='strict')
hdr = cutout.wcs.to_header()
print(hdr['CRVAL1'], hdr['CRVAL2'])
sys.exit()
hdr['BMAJ'], hdr['BMIN'], hdr['BPA'] = self.get_beam()
self.img_hdr = hdr
self.img_data = cutout.data
logging.info(f'{self.imagefile}: recenter, size ({shape[0]}, {shape[1]})' \
f' --> ({new_shape[0]}, {new_shape[1]})')
def new_HST_image_plot(storage,f, axloc):
img_file = '/Users/jansen/Google Drive/Astro/KMOS_SIN/HST_data/HST_'+storage['X-ray ID']+'.fits'
img=pyfits.getdata(img_file)
img_wcs= WCS(img_file).celestial
hdr=pyfits.getheader(img_file)
new_size = storage['Phys_size']
print (new_size)
try:
pixscale=abs(hdr['CD1_1']*3600)
except:
pixscale=abs(hdr['CDELT1']*3600)
print (pixscale)
Cat = storage['Cat_entry']
try:
Ra_opt = Cat['ra']
Dec_opt = Cat['dec']
except:
Ra_opt = Cat['RA']
Dec_opt = Cat['DEC']
opt_world= np.array([[Ra_opt,Dec_opt]])
opt_pixcrd = img_wcs.wcs_world2pix(opt_world, 0) # WCS transform
opt_x= (opt_pixcrd[0,0]) # X pixel
opt_y= (opt_pixcrd[0,1]) # Y pixel
position = np.array([opt_x, opt_y])
cutout = Cutout2D(img, position, new_size/pixscale, wcs=img_wcs,mode='partial')
img=(cutout.data).copy()
img_wcs=cutout.wcs
cut_opt = img
if type(axloc)==str:
ax = f.add_subplot(axloc, projection=img_wcs)
else:
ax = f.add_subplot(axloc[0],axloc[1],axloc[2], projection=img_wcs)
rms = np.std(cut_opt - cut_opt.mean(axis=0))
#rms = 0.13/3
print ('RMS of the HST', rms)
ax.imshow((cut_opt), origin='low', vmin=0, vmax=3*rms)
ax.set_autoscale_on(False)
return storage, ax
def cutout_exclusion_mask(self, fov='9 deg'):
"""Cutout appropriate part of exclusion mask
In many cases the exclusion mask is given as all-sky image, but only a
small fraction of that image is needed
Parameters
----------
exclusion : `~gammapy.image.ExclusionMask`
Input exclusion mask
fov : `~astropy.coordinates.Angle`
Field of view
"""
from astropy.nddata import Cutout2D
from astropy.nddata.utils import PartialOverlapError
fov = Angle(fov)
exclusion = self.exclusion
try:
c = Cutout2D(exclusion.mask,
self.on_region.pos,
fov,
exclusion.wcs,
copy=True,
mode='strict')
except PartialOverlapError:
raise PartialOverlapError('FOV ({}) not completely contained '
'in exclusion mask'.format(fov))
self.exclusion = ExclusionMask(name=exclusion.name,
data=c.data,
wcs=c.wcs)
def find_star(infile, pos=pos0, find_size2d=size2d):
## Mod on 24/02/2017 to handle CRs (quick fix)
## Mod on 01/04/2017 to handle CRs and bad pixels using cosmicrays_lacosmic
## Mod on 04/04/2017 to pass pos and find_size2d keywords
im0, hdr0 = fits.getdata(infile, header=True)
# + on 01/04/2017
im0_clean = cosmicray_lacosmic(im0, sigclip=10)
# Mod on 04/04/2017
# find_size2d = size2d #u.Quantity((25, 770), u.pixel)
cutout = Cutout2D(im0_clean[0],
pos,
find_size2d,
mode='partial',
fill_value=np.nan)
cutout = cutout.data
peak = np.where(cutout == np.max(cutout))
ycen, xcen = peak[0][0], peak[1][0]
# print xcen, ycen
xcen += pos[0] - find_size2d[1].value / 2.0
ycen += pos[1] - find_size2d[0].value / 2.0
# print xcen, ycen
return xcen, ycen
def test_cut_image_stamps():
setup = pipeline_setup.pipeline_setup({'red_dir': TEST_DATA})
stamp_dims = (20, 20)
image_file = os.path.join(TEST_DATA,
'lsc1m005-fl15-20170701-0144-e91_cropped.fits')
image = fits.getdata(image_file)
detected_sources_file = os.path.join(
TEST_DATA, 'lsc1m005-fl15-20170701-0144-e91_cropped_sources.txt')
detected_sources = catalog_utils.read_source_catalog(detected_sources_file)
stamps = psf.cut_image_stamps(setup, image, detected_sources[0:100, 1:3],
stamp_dims)
test_stamp = Cutout2D(np.ones([100, 100]), (50, 50), (10, 10))
got_stamp = False
for s in stamps:
if type(s) == type(test_stamp):
got_stamp = True
assert got_stamp == True