Home   GCX Home


Test images for aperture photometry evaluation


Radu Corlan     December 28, 2004

Download

test-images.zip     The test images and data tables.

Test Images

In order to verify the precision of aperture photometry routines and evaluate different algorithms, "life-like" star field images with precisely known flux values are needed. Four synthetic images were created from sequence files* of the AU CYG region.

The first two images:

aucyg-sparse-gaussian-2.5.fits
aucyg-sparse-moffat-2.5.fits

contain only stars that are brighter than the 15-th magnitude. Stars that were close together have been deleted. These images can be used to gauge the precision of the reduction routines in sparse field conditions (so they check for algorithm errors). The star positions and magnitudes are listed in "sparse-stars.tbl" (which is converted from the gcx star file sparse-stars.rcp).

The other two files:

aucyg-gaussian-2.5.fits
aucyg-moffat-2.5.fits

contain all the stars in the sequence file (down to mag 19 or so). The complete listing is in "aucyg-allstars.tbl" (which is converted from aucyg-allstars.rcp). They are useful to test the influence of contaminating objects.

The files contain synthetic stars images of either Gaussian or Moffat profiles, both with a FWHM of 2.5 pixels. The Moffat profiles have a "beta" parameter of 4. The plate scale is 1.5 arcseconds/pixel. The FITS header contains wcstools-like WCS information.

The flux of the generated stars is precisely that listed in the table (so a magnitude of e.g. 14.929 should read "14.929000"). Some small errors are introduced by 16-bit quantisation. For stars brighter than mag 13, these errors are below 0.0001 mags. For stars between 14.5 and 15, they can be up to 0.0005. The positions of the stars' centroids are accurate to better than 0.1 pixels.

Screenshots

The screenshots below show the "sparse" and "full" images with star symbold overlaid.

sparse screenshot crowded screenshot

GCX Results

The magnitudes of the stars listed in "sparse-stars.tbl" were extracted from the test images using gcx. Two central aperture shapes were used: the "whole pixels" shape, which sums the flux from whole pixels that are within the the specified aperture radius; and "irregular polygon", method that includes a fraction of the intensity of pixels that are located on the central aperture boundary. The name comes from the fact that the calculation methos used approximated the circular aperture shape with an irregular polygon. This method is also used in the IRAF phot routine.

For each aperture shapes, radiuses of 2, 3, 4, 5, 6, 7, 8 and 9 pixels were used. The standard deviation of residuals of the fit of extracted magnitudes to the "true" ones was calculated. The full results of the test are listed in "test.out". The "runtests" script was used to generate this file. The results are plotted below.

sparse field error graph


crowded field error graph

* AU CYG area sequence by Arne Henden