def compute_scales(input_images, scale_type, mask_key):
""" From the list of images, use the central third of the image to
calculate the scaling factor requested by user. """
lx, ly = fits.getdata(input_images[0]).shape
centre = [lx/3, ly/3, lx*2/3, ly*2/3] # min_x, min_y, max_x, max_y
scales = []
for image in input_images:
im = utils.read_image_with_mask(image, mask_keyword=mask_key, limits=centre)
if scale_type == "median":
new_item = numpy.ma.median(im)
elif scale_type == "mean":
new_item = numpy.ma.mean(im)
elif scale_type == "none":
new_item = 1
if new_item is numpy.ma.masked: # Result if all elements where masked
new_item = 1
print "All elements of the central part of image %s seem to be masked. "+\
"Replacing scale calculation by 1." % image
scales.append(new_item)
return scales # cases where of all elements were masked
def cube_images(input_images, mask_key, scales, limits=0):
""" From a set of images (input_images), and once divided by their scales,
form a cube with all the images. """
# Set the limits to build the piece of the cube
header = fits.getheader(input_images[0])
if limits == 0:
min_x, min_y = (0, 0)
max_x, max_y = header["NAXIS2"], header["NAXIS1"]
else:
min_x, min_y, max_x, max_y = limits
# Initialize the cube and the mask.
lx, ly = max_x - min_x, max_y - min_y
lz = len(input_images)
mask = numpy.zeros([lz,lx,ly]) # initialized to zero, all pixels are valid
cube = numpy.ma.array(mask, mask=mask)
# Now read all the images (and masks, if present), into the cube
for index, image in enumerate(input_images):
im = utils.read_image_with_mask(image, mask_keyword = mask_key, limits=[min_x, min_y, max_x, max_y])
cube.data[index,:,:] = im.data/scales[index]
cube.mask[index,:,:] = im.mask
return cube
def filter_image(args):
""" Routine that uses a median filter on a list of masked images. """
output_list = []
for image in args.input:
# Read image, mask and header
im = utils.read_image_with_mask(image, mask_keyword=args.mask_key)
hdr = fits.getheader(image)
# skimage uses masks where 1 means valid and 0 invalid
mask = (im.mask + 1) % 2
filt_im = skimages.filter.median_filter(im.data, mask=mask, radius=args.radius)
# Make name of file if not given
if args.output == "":
output = utils.add_suffix_prefix(image, suffix='-mf')
else :
output = args.output
# Add history line to the header and write to file
hdr.add_history("- Image median filtered. Radius = " + str(args.radius))
fits.writeto(output, filt_im, header=hdr)
output_list.append(output)
return output_list
def arith(args):
output_list = []
for image in args.input1:
# Read inputs as masked arrays or numbers (input2 might be a scalar)
im1 = utils.read_image_with_mask(image, args.mask_key)
try: # if second operand is a number we still need it in a masked array
im2 = numpy.ma.array([float(args.input2[0])],mask=[0])
except (ValueError,TypeError):
im2 = utils.read_image_with_mask(args.input2[0], args.mask_key)
# Do the actual operation. Result is a masked array which masks
# any pixel if it is masked either in im1 or in im2.
operations = {"+":operator.add,
"-":operator.sub,
"*":operator.mul,
"/":operator.div,
"**":operator.pow}
# Case of mean or median for Input2
if args.median == True:
operand2 = numpy.ma.median(im2)
elif args.mean == True:
operand2 = numpy.ma.mean(im2)
else:
operand2 = im2.data
oper = operations[args.operation[0]]
result = numpy.zeros_like(im1)
result.data[:] = oper(im1.data, operand2) # Actual operation of images
result.mask[:] = im1.mask | im2.mask # If any is masked, result is
# If args.fill_val is present, use it
if args.fill_val != '':
result.data[:] = result.filled(float(args.fill_val))
# If output exists use it, otherwise use the input. Prefix/suffix might
# modify things in next step.
if args.output != '':
outpt = os.path.abspath(args.output)
else:
outpt = os.path.abspath(image)
# Separate (path, file) and (file root, extensions). Then build new name.
outdir, outfile = os.path.split(outpt)
outfile_root, outfile_ext = re.match(r'(^.*?)(\..*)', outfile).groups()
outpt = os.path.join(outdir, args.prefix + outfile_root + args.suffix +
outfile_ext)
# Now name for the mask, if the name exists, use it, otherwise build it.
if args.mask_name != "":
mask_name = args.mask_name
else:
mask_name = outpt + ".msk"
# Prepare a header starting from input1 header
hdr_im = fits.getheader(image)
name2 = os.path.split(args.input2[0])[1]
if args.median: # if single number because median used
name2 = "median(" + name2 + ") (" + str(im2) + ")"
elif args.mean: # if single number because mean used
name2 = "mean(" + name2 + ") (" + str(im2) + ")"
# Write a HISTORY element to the header
#hdr_im.add_history(" - Operation performed: " + image + " " +
# args.operation[0] + " " + name2)
try:
hdr_mask = fits.PrimaryHDU(result.mask.astype(int)).header
except:
print type(im1), type(result), type(im2)
print im1
print result
print im2
hdr_mask.add_history("- Mask corresponding to image: " + outpt)
# Now save the resulting image and mask
if os.path.isfile(outpt):
os.remove(outpt)
if os.path.isfile(mask_name):
os.remove(mask_name)
fits.writeto(outpt, result.data, header=hdr_im)
fits.writeto(mask_name, result.mask.astype(numpy.int0), header=hdr_mask)
output_list.append(outpt)
return output_list
def arith(args):
output_list = []
for operand1, operand2 in zip(args.input1, args.input2):
# Read inputs as masked arrays or numbers (input2 might be a scalar)
im1 = utils.read_image_with_mask(operand1, args.mask_key)
try: # if second operand is a number we still need it in a masked array
value2 = numpy.ma.array([float(operand2)],mask=[0])
except (ValueError,TypeError):
value2 = utils.read_image_with_mask(operand2, args.mask_key)
# If --keyword was passed, read from the header of the image the value and use it, instead of the full image
if args.keyword:
value2 = numpy.ma.array( fits.getval(operand2, args.keyword) )
# Do the actual operation. Result is a masked array which masks
# any pixel if it is masked either in im1 or in im2.
operations = {"+":operator.add,
"-":operator.sub,
"*":operator.mul,
"/":operator.div,
"**":operator.pow}
# Case of mean or median for Input2
if args.median == True:
value2 = numpy.ma.median(value2)
elif args.mean == True:
value2 = numpy.ma.mean(value2)
oper = operations[args.operation[0]]
result = numpy.zeros_like(im1)
result.data[:] = oper(im1.data, value2.data) # Actual operation of images
result.mask[:] = im1.mask | value2.mask # If any is masked, result is masked as well
# If args.fill_val is present, use it
if args.fill_val:
result.data[:] = result.filled(float(args.fill_val))
# If output exists use it, otherwise use the input. Prefix/suffix might
# modify things in next step.
if args.output != '':
outpt = os.path.abspath(args.output)
else:
outpt = os.path.abspath(operand1)
# Separate (path, file) and (file root, extensions). Then build new name.
outdir, outfile = os.path.split(outpt)
outfile_root, outfile_ext = re.match(r'(^.*?)(\..*)', outfile).groups()
outpt = os.path.join(outdir, args.prefix + outfile_root + args.suffix +
outfile_ext)
# Now name for the mask, if the name exists, use it, otherwise build it.
if args.mask_name != "":
mask_name = args.mask_name
else:
mask_name = outpt + ".msk"
# Prepare a header starting from input1 header
hdr_im = fits.getheader(operand1)
name2 = os.path.split(operand2)[1]
if args.median: # if single number because median used
name2 = "median(" + name2 + ") (" + str(value2) + ")"
elif args.mean: # if single number because mean used
name2 = "mean(" + name2 + ") (" + str(value) + ")"
# Write a HISTORY element to the header
#hdr_im.add_history(" - Operation performed: " + image1 + " " +
# args.operation[0] + " " + name2)
try:
hdr_mask = fits.PrimaryHDU(result.mask.astype(int)).header
except:
pass
hdr_mask.add_history("- Mask corresponding to image: " + outpt)
# Update the keyword if asked by user
if args.update and args.keyword:
old_keyword = hdr_im[args.keyword]
hdr_im[args.keyword] = (oper(old_keyword, value2.data), "Before: {0}".format(old_keyword) )
# Now save the resulting image and mask
if os.path.isfile(outpt):
os.remove(outpt)
if os.path.isfile(mask_name):
os.remove(mask_name)
fits.writeto(outpt, result.data, header=hdr_im)
fits.writeto(mask_name, result.mask.astype(numpy.int0), header=hdr_mask)
utils.header_update_keyword(outpt, "MASK", os.path.abspath(mask_name),
comment="Name of mask image. ")
output_list.append(outpt)
return output_list