nvieira-mcgill / amakihi

End-to-end image differencing and transient detection software

amakihi

End-to-end image differencing and transient detection software, including:

• Downloading reference images (a.k.a. templates) from various surveys
• Image background subtraction/estimation
• Image masking, including masking of saturated pixels
• Building effective Point-Spread Functions (ePSFs)
• Image alignment (a.k.a. image registration)
• Image differencing (a.k.a. image subtraction)
• Transient detection and basic vetting of candidate transients

The end products of a pipeline constructed from amakihi are "triplets", i.e., N length-3 arrays of the (science, reference, difference) images cropped around N candidate transient sources. These candidate transients can then be further vetted with e.g. your favourite machine learning algorithm. (I use braai).

Some modules for interfacing with braai are included here as well.

This software was developed to serve in a pipeline for the MegaCam instrument of the Canada-France-Hawaii Telescope (CFHT). This pipeline was used for all image differencing and transient detection in the following paper describing our CFHT MegaCam follow-up of the gravitational wave event GW190814:

Vieira, N., Ruan, J.J, Haggard, D., Drout, M.R. et al. 2020, ApJ, 895, 96, 2. *A Deep CFHT Optical Search for a Counterpart to the Possible Neutron Star - Black Hole Merger GW190814.

Documentation

Detailed documentation for all modules can be found here.. In the future, example scripts/notebooks will be added.

Installation

Currently, needs to be installed directly from github. May be installable with conda and/or pip in the future.

Dependencies must be installed manually for the time being. Dependencies are:

Essential dependencies:

• numpy
• scipy
• matplotlib
• astropy
• photutils
• scikit-image
• scikit-learn
• astroquery (for crossmatch module)
• lmxl (for query_CFIS module)

Semi-optional dependencies:

• tensorflow (if using rb_model_utils.write_model(), .load_model(), .use_model(), or .vgg6())

Really optional dependencies:

• imblearn (if using rb_dataset_utils.augment_dataset_SMOTE())

Non-Python:

• astrometry.net (can however be ignored in favour of source detection with the image segmentation methods of photutils instead)
• hotpants

Once you have the dependencies, to install amakihi, simply run :

python setup.py install

Contact

[e-mail]

Credits

Free software: MIT license

This package was created in part with Cookiecutter and the audreyr/cookiecutter-pypackage project template.