I'm generally unhappy with the values that my reduction pipeline is producing.
This page will document what I'm doing and allow others to look at what I've
developed. Also I tend to find my own problems when I look closely and document
my work. All image calculations presented below have been verified with
AIP4Win as appropriate.
All fits file interactions are accomplished with
The first step in the reductions is to create master dark and flat images.
A perl script called masterCreate.pl analyzes the available
dark images and feeds the image list to darkCreate to munge into an
appropriate image. The same happens for flat images using flatCreate,
except star images are
used as I currently have no ability to take true flat images.
A utility tool I use to calculate image statistics called 'stdimg' was checked
against 'fimgstat'. An image was chosen at random and the data given by
stdimg 3050/V_flat.fits was compared against
fimgstat 3050/V_flat.fits INDEF INDEF, and no differences
Image selection stuff.
Using fimgcreate 16 "2064,2034" "none" zero.fits,
an image of appripriate
size was created filled with 0's. Then, five more images were created using
farith zero.fits 1 one.fits +,
farith zero.fits 2 two.fits +,
farith zero.fits 3 three.fits +,
farith zero.fits 4 four.fits +,
farith zero.fits 5 five.fits +.
Testing was done after darkCreate.c
was fixed and the correct results were achived for both image sets (2.5 and 3)
and all three methods (mean, median and IQM).
First, three runs were executed using darkCreate, each using either the mean,
median or IQM (Inner Quartile Mean) with the images one.fits, two.fits,
three.fits and four.fits.
fimgstat ?.fits INDEF INDEF was then
executed on each of the images generated, which all produced min/max/mean
values of 2.5, which is appropriate for the four images given.
three runs were executed using darkCreate, each using either the mean,
median or IQM with the images one.fits, two.fits, three.fits, four.fits and
five.fits. The min/max/mean produced values of 3, 4 and 2.5 respectivly, which
is wrong for median and IQM. The median problem was off by one image when
an odd number of images was used.
The IQM problem was not using the last image
when there were an odd number of images.
The program flatCreate.c suffered from the same problem as darkCreate.c, and
the same fix was applied.
DAOPHOT options and FWHM usage.
How matched against TYCHO-2.
Changed from an iterative Perl script to a
GSL C program.