Skip to main content

Linear Feature Detector for Astronomical images

Project description

Linear Feature Detector


Linear Feature Detector (LFD) library is a collection of packages that enable user to detect and analyze linear features on astronomical images. The code was designed to be run interactively, or at scale on a cluster and specifically targets SDSS survey data and meteor trails therein.

LFD is a more complete version of LFDS that contains all of the never-published features of LFDS. Except for the linear feature detection code in the detecttrails module, most of the LFDS code was recoded from scratch and made compatible with Python 3 and OpenCv 3.0. You can find the old LFDS code here_.

.. _here:


Install from pip by running

.. code-block:: bash

pip install lfd

or clone it locally and use requirements.txt to create an environment from which you can run lfd

.. code-block:: bash

git clone

Import lfd and be on your merry way. One issue can occur when using conda or miniconda virtual environments where numpy can not be found. In that case it is neccessary to run conda install numpy and repeat the pip install for the lfd to install properly.


Major requirements are as follows

  • Python 3+
  • OpenCV 3+
  • NumPy 1.11+
  • SciPy 0.19+
  • Fitsio 0.9.7+
  • SQLAlchemy 1.2.11+
  • parts of Erin Sheldon's esutil_ and sdsspy_ utilities are bundled with the provided code. Some of the code might have been altered

.. _esutil: .. _sdsspy:

Running the code

Read the docs! They contain many examples.

By default lfd is setup to work with SDSS files and directory structure. This can be altered significantly, although complete departure from SDSS file and directory structures are not supported out of the box.

Although slightly out of data much of the processing steps are still adequatly described in::

Bektesevic & Vinkovic, 2017, MNRAS, 1612.04748, Linear Feature Detection Algorithm for Astronomical Surveys - I. Algorithm description

To start processing use any of the following:

.. code-block:: python

 import lfd

 foo = lfd.detecttrails.DetectTrails(run=2888)
 foo = lfd.detecttrails.DetectTrails(run=2888, camcol=1)
 foo = lfd.detecttrails.DetectTrails(run=2888, camcol=1, filter='i')
 foo = lfd.detecttrails.DetectTrails(run=2888, camcol=1, filter='i', field=139)

It is possible to change detection parameters of any step in the processing by

.. code-block:: python

 foo.params_bright["debug"] = True
 foo.params_removestars["filter_caps"]["i"] = 20

Results are outputted to a file provided by the filepath results, by default set to results.txt. Results file is a CSV file in which the detected parameters. Results module provides functionality to parse these CSV files into a database for which an SQLAlchemy ORM is provided.

.. code-block:: python

 from lfd.results import Event, Frame, Point
 from lfd import results

 # create or connect to a database

 # populate it with data either from output of detecttrails

 # or create mock data to play with

 # query on Event or Frame parameters fo a single or a collection of items
 with results.session_scope() as s:
     # returns all Events found on run 2888, but pick only one
     e = s.query(Event).filter(

     # get a collection of frames 
     fquery = query.filter(Frame.t.iso > '2009-09-27 10:06:10.430')
     f = fquery.all()
     lfd.results.deep_expunge_all(f, s)

 # create table like output

 # manipulate them as OO objects and commit the changes back, f.e. move one
 # of the points of the line somewhere else
 e.p1 = Point(10, 10, camcol=5, filter='r')

 # or just move one of P1(x1, y1), P2(x2, y2) line coordinates
 e.y2 = 10

 # see and work with the coordinates values in reference to the origin of
 # the entire CCD array and not just individual CCDs within

 # equivalent to
 e.cx1 = 100

 # find the points where the line corsses the individual CCD edges again and go there

 # persist the changes to the DB
 with results.session_scope() as s:

LFD was designed to be able to handle processing large ammounts of data, in fact it was used to process the entire SDSS database of images by using the Fermi cluster at Astronomical Observatory Belgrade in Serbia. To make the creation of scripts that ran LFD on the cluster easier createjobs module was written. By default it is oriented towards running on that particular cluster, but it should be easily adaptable to any Sun Grid cluster out there.

.. code-block:: python

 jobs = cj.Jobs(500)
 There are no runs to create jobs from.
   Creating jobs for all runs in runlist.par file.

   765 jobs with 1 runs per job
   Queue:     standard
 Wallclock: 24:00:00
 Cputime:   48:00:00
 Ppn:       3
 Path:      /home/user/Desktop/.../jobs

which is of course very flexible

.. code-block:: python

runs = [125, 99, 2888, 1447] cmd = """python3 -c "import detecttrails as dt; x = dt.DetectTrails($); x.params_bright['debug']=True; x.process()""" jobs = cj.Jobs(2, runs=runs, camcol=1, filter='i', command=cmd) jobs.create()

User will be notified about all important parameters that were set. LFD also comes with Graphical User Interfaces through which users can create these jobs via mouseclicks but also visually inspect their results by using the provided specially designed image browser.

An analysis module is provided as well through which theoretical meteor profiles can be generated as described in::

Bektesevic & Vinkovic et. al. 2017 (arxiv: 1707.07223).

.. code-block:: python

 from lfd.analysis import profiles

 point = profiles.PointSource(100)
 seeing = profiles.GausKolmogorov(profiles.SDSSSEEING)
 defocus = profiles.FluxPerAngle(100, *profiles.SDSS)

 a = profiles.convolve(point, seeing, defocus)

 import matplotlib.pyplot as plt
 fig, ax = plt.subplots(1, 1)
 profiles.plot_profiles(ax, (point, seeing, defocus, a))

All of this is, of course, just a quick overview of all functionalities. There are many more details describing this and other useful utilities, including Graphical User Interfaces to common functionality, provided by LFD availible in the documentation.


GNU GPLv3 Copyright (C) 2018 Dino Bektesevic

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see

.. __licenses:

.. |docs| image:: :alt: Documentation Status :scale: 100% :target:

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for lfd, version 2.0.0
Filename, size File type Python version Upload date Hashes
Filename, size lfd-2.0.0-py3-none-any.whl (1.9 MB) File type Wheel Python version py3 Upload date Hashes View
Filename, size lfd-2.0.0.tar.gz (1.9 MB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring DigiCert DigiCert EV certificate Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page