===================
- Avoid applying the estimated cosmic ray vs real sources threshold for the
ACS/SBC and WFC3/IR detectors. [1858]
- Corrected the way the n1_exposure_time and tot_exposure_time values
are computed as these values are used in the computation for rejecting
catalog creation based on expected cosmic ray detections. Generalized
the crfactor dictionary for all detectors. Ensure if any catalog type
is rejected, all the catalog types are rejected. [1853]
- Modified the call to the hamming function in the deconvolve_utils.py module
as SciPy deprecated the way window filtering functions can be invoked. These
functions can no longer be imported from the scipy.signal namespace but need
to be accessed via scipy.signal.windows. [1848]
- Corrected the way that the number of constituent images are accumulated
per pixel by ensuring each contributing pixel has a finite value and
is not zero. [1820]
- Within the HAP configuration files, increased the minimum number of matches
for a successful "rscale" fit from 6 to 10, and removed "shift" as a fit geometry
option. [1823].
- Removed the use of a custom smoothing kernel based upon actual image
data as a poorly determined kernel can ultimately cause poor source
position determination. The default kernel has been set to a
Gaussian with default dimensions of 11 x 11 pixels. [1805]
- Addressed bugs caught by SonarQube static code analysis. Interface
changes listed here: Added missing input data parameter to the create_output
calls, Added missing log level to run function, Removed the deprecated
parameter, dao_threshold, from astrometric_utils.py/extract_sources, removed
"ivmlist" parameter from the interface of multiple functions in processInput.py
as it is an output parameter (buildFileListOrig, buildFileList, checkMultipleFiles,
and process_input), and addressed missing parameters in the calls to
get_ci_info and get_ci_from_file.. [1802]
- Exclude single filter images from the generation of the total detection
image to minimize cosmic ray contamination, unless there are only single
filter images in the visit. [1797]
- Implemented a series of bug fixes for the segmentation catalog [1793]
- Define the threshold image to be (nsigma * background_rms).
- Fixed bug in the generation of the threshold image - ensure the final
threshold is built up properly by using the weight mask for the region
in question.
- Pass the background image to detect_segments() so the convolved image can be
background subtracted.
- For the detection of sources, background subtract the input image for both the
Gaussian and RickerWavelet kernels. Do not do any clipping on the background
subtracted image.
- Update configuration files for the RickerWavelet2DKernel: source_box is now 6
and rw2d_nsigma is now 3.
- Fixed a bug in the computation of the "biggest source".
- Created a new method, ricker_matched_kernel(), to generate the RickerWavelet2DKernel
properly. Sigma is now provided, versus the FWHM, to the RickerWavelet2dKernel
constructor, and the normalization is handled by the new method where the
normalization causes the RickerWavelet core to match the Gaussian core. [1791]
- Added contributors guide to readthedocs. [1787]
- Removed "tophat" as a kernel option, added warnings for "gaussian" and "lanczos3"
that they may not be conserving flux. [1786]
- Updated config json to exclude bad pixels in single WFC3/IR SVM processing. [1783]
- Bug fix for mdriztab=True option in Astrodrizzle previously overwriting user inputs. [1774]
- Reverted PR 1222 allowing pixels to be filled with available data where WHT=0. [1767]
- Force the identified bad rows to be removed from the total (aka white light)
source catalog before the corresponding bad segments are removed from the
segmentation image. [1771]
- Improved calculation of S_REGION using dialation and erosion. [1762]
- Skycell added to flt(c) and drz(c) science headers for the pipeline and svm products. [1729]