| Publications | |
Back to Stefan's home page
Licentiate thesis by Stefan Spännare, Lund Observatory, January 1997
The program POLYFIT has been developed to perform photometry of well
sampled stars on luminous uneven backgrounds consisting of nebulae or
galaxies. The program can also be used to remove stars from galaxy
images, if the galaxy itself is the relevant object. The program uses
empirical PSFs defined from the stellar image. POLYFIT performs
a simultaneous least squares fit of the PSF to the image of the star
and a two-dimensional polynomial (of degree ≤ 5) to the background.
Detection and approximate positions of the stars in the image are made
available by a separate routine. As a rule of thumb a fitting radius of
1.5·FWHM of the image of the star and a polynomial degree of 2 or 3 give
the best results. Comparative tests using simulated and observed well
sampled stellar images added to observed galaxy backgrounds show that
POLYFIT gives photometric results superior to those produced by some
major photometry packages widely available. Photometric errors smaller
than 0.05 magnitudes are obtained for most stars in these images.
Download the thesis (4 files):
| lic1.pdf 192 kB | lic1.ps.gz 233 kB | Cover page, 1 figure |
| lic2.pdf 16 kB | lic2.ps.gz 11 kB | Photometry overview |
| lic3.pdf 116 kB | lic3.ps.gz 85 kB | Thesis text, 1 figure |
| lic4.pdf 1899 kB | lic4.ps.gz 2379 kB | Thesis figures |
P. Linde and S. Spännare, Lund Observatory, June 1993
A local implementation of the Lucy-Richardson algorithm has been used
to deconvolve a set of crowded stellar field images. The effects
of deconvolution on detection limits as well as on photometric and
astrometric properties have been investigated as function of the number
of deconvolution iterations. Results show that deconvolution improves
detection of faint stars, although artifacts are also found. Both
automatic and interactive detection methods benefited. However, the
interactive method always gave the better results. Regarding astrometry,
deconvolution provides more stars measurable without significant
degradation of positional accuracy. The photometric precision is affected
by deconvolution in several ways. Errors due to unresolved images are
notably reduced while flux redistribution between stars and background
increases the errors.
Download the article (1 file):
| article1.pdf 325 kB | article1.ps.gz 276 kB | Complete article |