| Author: | Brendt Wohlberg <brendt@ieee.org> |
|---|
This directory contains the scripts used to generate the results in the paper PSF Estimation in Crowded Astronomical Imagery as a Convolutional Dictionary Learning Problem (doi:10.1109/LSP.2021.3050706). The proposed method, referred to here as Interpolated Convolutional Dictionary Learning (ICDL), is compared with Resolved Component Analysis (RCA).
The implementation of the ICDL method requires SPORCO version 0.2.0 or later. It can be installed by
pip install sporco
or, within a conda environment
conda install sporco
See the file INSTALL_RCA.rst for instructions on installing RCA and its dependencies.
- compute_rca_opt_param.py
- Find optimal parameters for each test case by optimizing over 9000 different parameter combinations. The RCA scripts below depend on the results of this script, but pre-computed results are included in this repository, so it is not essential to re-run this script.
- tabulate_rca_opt_param.py
- Tabulate the results computed by compute_rca_opt_param.py.
- compute_rca_results.py
- Compute RCA solutions for each test case using optimal parameters found by compute_rca_opt_param.py.
- tabulate_rca_results.py
- Tabulate the results computed by compute_rca_results.py.
- compute_icdl_results.py
- Compute ICDL solutions using parameter selection heuristics.
- tabulate_icdl_results.py
- Tabulate the results computed by compute_icdl_results.py.
- compute_icdl_opt_param.py
- Find optimal parameters for each test case by optimizing over 768 different parameter combinations. These results are not included in the above-mentioned paper. Pre-computed results are included in this repository, so it is not essential to re-run this script before running tabulate_icdl_opt_param.py.
- tabulate_icdl_opt_param.py
- Tabulate the results computed by compute_icdl_opt_param.py.
- make_psf_plots.py
- Construct PSF estimation performance comparisons included in the extended version of the paper