[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

New version of xrt_prep software and associated programs

hi everyone,

Apologies if you get multiple versions of this message.
I've just completed a major upgrade to <xrt_prep.pro>.  There are a
couple of new related programs as well:

<xrt_med_dark.pro>  - generates a filtered, medianed dark from temporally
nearby darks.

<xrt_pixel_grade.pro> - generates and/or "interprets"  a coded pixel "grade"
array, indicating the locations (and fractional occupancy for binned data)
of saturated, saturation bleed, contamination spot, dust and "hot" pixels.
This should be useful as a stand alone program as well (in "interpretation"

These programs have been added to the SSW tree locally at SAO and the
new version of xrt_prep is the new default.  They should appear in the
worldwide SSW tree in the next few days.

There are several changes:

(1) The dark correction has been improved.  The zero-point is now
better determined with lower scatter, and the shape of the dark is
better estimated.  This comes at a cost of longer processing time
- 5 full-frame darks nearest in time to the data being prepped are
gathered, filtered and medianed to "reset" the zero-level (using
the new <xrt_med_dark.pro>).  This generally sets a minimum processing
time of ~10 s per image, with a maximum of ~30s for a 2048^2 image.
The user can optionally return to the old default (no median-ed
dark adjustment, faster run time, but lower zero-point accuracy)
by setting a new keyword dark_type=2, or they may use just the
median-ed darks as the dark frame (by setting dark_type=1).

(2) The old saturation array has been replaced by a new "pixel-grade" array,
grade_map (generated by the new <xrt_pixel_grade.pro>).  It encodes the
positions of not only

a) saturated pixels,
but also
b) saturation bleed pixels (ie, DN<2500 but affected by nearby saturation)
c) contamination spot pixels
d) dust speck pixels
e) "hot" pixels

In binned data, grade_map can optionally have a fraction attached which
further encodes the number of 1x1 pixels of types c,d, & e in each 
binned pixel.
The old optional sat_map keyword parameter is replaced by grade_map,
with the associated n_sat_pixels replaced by n_grade_pixels.  The latter
is now 2 dimensional, reflecting the number of each of the 5 pixel grades.
A single output n1xn2 grade_map array may be disassembled an n1xn2x5 array
grade_maps (ie, one map per grade type) using the new program
<xrt_pixel_grade.pro>, e.g.,

xrt_pixel_grade, index, grade_map, grade_maps, interpret=1

if fractional occupancies are wanted (in the case of binned data):

xrt_pixel_grade, index, grade_map, grade_maps, interpret=1, grade_frac=1

see header info on <xrt_pixel_grade.pro> for more details.

(3) there have been some minor improvements to the nyquist (odd-even bias)
removal (<no_nyquist.pro>), and to the model dark subtraction
(<lsback_away.pro>).  There will probably be some further incremental
improvements to both in the not-to-distant future.

(4) There is added ability to adjust the Fourier filtering at the top
level (clean_type, nsigma, and nmed keywords).  See the program header
for details.

When in doubt, read the program header.  Hope you like
the results & let us know if you find any problems!


steve saar