[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Precise centroid calculation and WCS-SIP fitting
From: |
Mohammad Akhlaghi |
Subject: |
Re: Precise centroid calculation and WCS-SIP fitting |
Date: |
Fri, 27 Oct 2023 14:53:32 +0200 |
User-agent: |
Mozilla Thunderbird |
Dear Rafael,
The centroids can be found using MakeCatalog, either in pixel
coordinates (first link below) or WCS coordinates (second link):
https://www.gnu.org/software/gnuastro/manual/html_node/Position-measurements-in-pixels.html
https://www.gnu.org/software/gnuastro/manual/html_node/Position-measurements-in-WCS.html
MakeCatalog needs two input images for its measurements: a "values"
image (this is usually the scientific image and is usually in floating
point type) and a "label" image (that has an integer type and each group
of pixels that are meant to be measured together are given the same
integer identifier). This is fully described in the documentation of
MakeCatalog:
https://www.gnu.org/software/gnuastro/manual/html_node/MakeCatalog.html
The usual way is that you first run NoiseChisel to detect signal, then
you run Segment to identify separate peaks within the detections and
finally you run MakeCatalog with your reqested measurements like below:
astnoisechisel image.fits --output=nc.fits
astsegment nc.fits --output=seg.fits
astmkcatalog seg.fits --clumpscat --ids --ra --dec --x --y \
--output=catalog.fits
If the peaks are important for you and not the diffuse regions, you want
the 'CLUMPS' catalog (second HDU of 'catalog.fits' in the example
above), not the 'OBJECTS' catalog (first HDU of 'catalog.fits'):
asttable catalog.fits -hCLUMPS
But of course, as described at the beginning of the MakeCatalog
documentation above, you can build your labeled images in any way you
prefer.
There is a full/complete tutorial on how to optimize NoiseChisel and
Segment and work with the output catalogs in several scenarios:
https://www.gnu.org/software/gnuastro/manual/html_node/General-program-usage-tutorial.html
I hope this helps,
Cheers,
Mohammad