TomoPhantom [1] is a toolbox to generate customisable 2D-4D phantoms (with a temporal capability) and their analytical projection data for various image processing tasks (reconstruction, denoising, deblurring, etc.).
TomoPhantom is recommended for various image processing tasks that require extensive numerical testing: image reconstruction, denoising, deblurring, etc. Specifically,TomoPhantom is best-suited for testing various tomographic image reconstruction (TIR) methods. For TIR algorithms testing, the popular Shepp-Logan phantom is not always a good choice due to its piecewise-constant nature. This toolbox provides a simple modular approach to efficently build customisable 2D-4D phantoms consisting of piecewise-constant, piecewise-smooth, and smooth analytical objects. The objects include: Gaussians, parabolas, ellipses, cones, rectangulars. The exact tomographic projections (analytical sinograms) can be also obtained. The sinograms can be used for TIR benchmarking purposes, which enables reconstruction without so-called 'Inverse Crime'. TomoPhantom is also compatable with ASTRA-toolbox and TomoPy packages. Generated data can be directly reconstructed using the toolboxes (see examples). Additionally, TomoPhantom provides a simple temporal extension, i.e. a capability of creating 2D+time and 3D+time models.
conda build conda-recipe --numpy 1.12 --python 3.5
conda install tomophantom --use-local --force
run compile_mex_linux.m % to compile CPU modules on linux
run compile_mex_windows.m % to compile CPU modules on Windows
- Phantom2DLibrary.dat and Phantom3DLibrary.dat are editable text files with parametrised models (2D/3D versions of Shepp-Logan, Defrise, and QRM phantoms are included). The generation of new phantoms is highly encouraged, please submit them through pull requests or via e-mail bellow.
- See MATLAB and Python demos
- TomoPhantom is released under Apache License v.2. Note that some demos where 'ASTRA-toolbox' is used are of GPLv3 license and also BSD-3 license for TomoPy package.
- xdesign XDesign is an open-source Python package for generating configurable simulation phantoms for benchmarking tomographic image reconstruction.
- syris Syris (synchrotron radiation imaging simulation) is a framework for simulations of X-ray absorption and phase contrast dynamic imaging experiments, like time-resolved radiography, tomography or laminography.
[2] D. Kazantsev, V. Pickalov "New iterative reconstruction methods for fan-beam tomography", IPSE, 2017
- Regularised FISTA-type iterative reconstruction algorithm for X-ray tomographic reconstruction with highly inaccurate measurements
- Joint image reconstruction method with correlative multi-channel prior for X-ray spectral computed tomography
Software related questions/comments please e-mail to Daniil Kazantsev at [email protected]