/spire

Simple Prototyping of Image REconstruction methods

Primary LanguagePython

Presentation

SPIRE (Simple Prototyping of Image REconstruction methods) is a python package providing wrappers and utilities for tomography and image processing algorithms.

It is divided in several parts:

  • The tomography submodule provides a simple wrapper of the ASTRA toolbox for parallel beam geometry. It also provides sinogram pre-processing (denoising, rings artifacts correction) and dataset exploration utilities.
  • The operators submodule implements various linear operators: FFT, Wavelet Transform, Convolution, tomographic projector, and other image operators.
  • The algorithms submodule provides proximal algorithms (FISTA, Chambolle-Pock) for minimizing an objective function involving a linear operator.

Installation

Installation from pypi

spire is available on the Python package index:

pip install --user spire

Installation from the sources

You can build and install this package from the sources.

git clone git://github.com/pierrepaleo/spire
  • You can directly install from the sources with
python setup.py install --user

Dependencies

To use SPIRE, you must have Python > 2.7 and numpy >= 1.8. These should come with standard Linux distributions.

Numpy is the only component absolutely required for SPIRE. For special applications, the following are required :

  • The ASTRA toolbox for tomography applications

  • pywt for Wavelets applications. This is a python module which can be installed with apt-get install python-pywt

Note : This module should be compatible with Python 3. However, all the dependencies must have a Python 3 version installed.

Documentation

A documentation is available in the doc/ folder.

Disclaimer

This module is at an early development stage. This means that the API may change at any time.