Photometry on images in Multi-Extension FITS (MEF) files

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

Photometry on images in Multi-Extension FITS (MEF) files

This post was updated on .
Multi-Extension FITS (MEF) files containing multiple images from detector arrays are not directly compatible with AIJ. However, the image array can be assembled into a single image FITS file using several methods.

The ideal method for Linux installations (and possibly OS X??) is to use swarp to assemble all of the MEF images into a mosaic which preserves the detector spacing and arrangement and results in a single image with valid WCS headers. swarp is available for download here:

An example command line would be:

swarp my_mef_file.fits -c swarp.default

where swarp.default is a good start at appropriate settings for the task at hand. The swarp.default file is available here to help users get started: swarp.default

If tracking and guiding are good for a time series of images, WCS headers are likely not required for apertures to track appropriate stars throughout the time series. In that case, a simple IDL/GDL or python script can be used to assemble the array of images into a single image, but because the images are edge-butted into an N x M array without accounting for the detector separation, a WCS solution is generally not possible on the resulting mosaic.

AIJ users have written and graciously provided IDL/GDL and python scripts that will assemble multiple image arrays into a single image FITS image that is compatible with AIJ, but that does not account for the detector spacing. Again, if good tracking/guiding is maintained throughout the image sequence, differential photometry can likely be run on the composite images without proper spacing or WCS headers. If you understand your detector spacing precisely, you may be able to adjust the scripts to add the spacing into the mosaic and a WCS solution may then be possible.

The first option requires IDL or the open source version GDL to run the script to convert a batch of images. The author offers this script as is and can't directly support it. But if you have questions, maybe other users could help out.

This particular version of the script assembles the image extension [1] in the lower left, [2] in lower right, [3] in upper left, and [4] in upper right. That ordering may need to be changed depending on your detector array orientation and detector control software. In this case, the four detectors in the array produce square 2046x2046 pixel images and the resulting single image is then a 4092 x 4092 pixel image. The user will likely need to customize this procedure to handle their specific multi-detector array configuration.

The IDL/GDL code is as follows:

pro convert_mef

imagedir = "/yourpath/"

naxis1 = 2046
naxis2 = 2046
readcol, ''+imagedir+'raw/image.list', imagename, format = 'a'
filename = imagedir + 'raw/' + imagename
nim = n_elements(filename)

;loop through the list of image files
for n=0,nim-1 do begin          

;combine each extension into one image
rdfits_struct,filename[n],quadi, /silent

im1 = quadi.im1
im2 = quadi.im2
im3 = quadi.im3
im4 = quadi.im4

image1 = im1[0:naxis1-1,0:naxis2-1]
image2 = im2[0:naxis1-1,0:naxis2-1]
image3 = im3[0:naxis1-1,0:naxis2-1]
image4 = im4[0:naxis1-1,0:naxis2-1]

image = dblarr(naxis1*2,naxis2*2)

image[0:naxis1-1,0:naxis2-1] = image1
image[0:naxis1-1,naxis2:naxis2*2-1] = image3
image[naxis1:naxis1*2-1,0:naxis2-1] = image2
image[naxis1:naxis1*2-1,naxis2:naxis2*2-1] = image4

header = headfits(filename[n])

;save processed image
cimagename = strmid(imagename[n], 0, 22)

writefits, imagedir+'pipeline/'+cimagename,image, header





The following python script provides the same functionality. Again, the script will likely need to be customized for your specific multi-detector configuration, but this should serve a good template for python users:


# Extract images from a FITS Multi-Extension file

import os
import sys
import numpy as np
import pyfits
from time import gmtime, strftime  # for utc

# Set output image size
oxsize = 4096
oysize = 4096

# Set input image size
ixsize = 2048
iysize = 2048

# Create a numpy array with zeros

outimage = np.zeros((oxsize,oysize))

# Set verbose flag True to show headers  and other information

verbose = True

# Set a clobber flag True so that images can be overwritten
# Otherwise set it False for safety

clobberflag = True  

if len(sys.argv) == 1:
  print " "
  print "Usage: meffile.fits"
  print " "
  sys.exit("Mosaic images from a fits Multi-Extension file\n")
elif len(sys.argv) == 2:
  infile = sys.argv[1]
  print " "
  print "Usage: meffile.fits"
  print " "
  sys.exit("Mosaic images from a fits Multi-Extension file\n")

# Extract the base name of the MEF file

filebase = os.path.splitext(os.path.basename(infile))[0]

# Open the fits file readonly by default and create an input hdulist

inlist =

nimages = len(inlist) - 1
if (nimages < 1):
  print "This file does not appear to be a FITS MEF file."

if verbose:

  print "The file "+infile+" contains %d images: \n" %(nimages,)
  print ""

  print "with the global header:"
  print inlist[0].header
  print ""

# Merge  the extension images into one numpy array
# The pattern here depends on how the images were placed in the MEF

outimage[0:ixsize-2,0:iysize-2] = inlist[1].data
outimage[0:ixsize-2,iysize:oysize-2] = inlist[2].data
outimage[ixsize:oxsize-2,0:iysize-2] = inlist[3].data
outimage[ixsize:oxsize-2,iysize:oysize-2] = inlist[4].data

# Write the array as a fits file

file_time = strftime("%Y-%m-%d %H:%M:%S", gmtime())
outlist = pyfits.PrimaryHDU(outimage)
outfile = filebase+"_mosaic.fits"
outheader = outlist.header
outheader['DATE'] = file_time
outheader['history'] = 'Created by'
outheader['history'] = '  from '+infile
outlist = pyfits.PrimaryHDU(outimage,outheader)
outlist.writeto(outfile, clobber = clobberflag)
if verbose:
  print "Combined Images written as %s \n" %(outfile,)