astra-toolbox/astra-toolbox ↗
Created Dec 27, 2020 · View the astra-toolbox/astra-toolbox repository page
ASTRA Tomography Toolbox
Want this for your repo?
Render a free sample of any GitHub repo in seconds.
Contributors
20
Lines of Code
6,193
From
Jul 1, 2013
To
Nov 13, 2020
About astra-toolbox/astra-toolbox
The ASTRA Toolbox is a high-performance GPU-accelerated software library for tomographic reconstruction and image processing, available as both Python and MATLAB toolboxes. It provides optimized primitives for forward and backward projection operations in 2D and 3D tomography, supporting multiple beam geometries including parallel beam, fan beam, and cone beam configurations with flexible source and detector positioning.
The toolbox includes implementations of numerous reconstruction algorithms such as FBP (Filtered Back Projection), SIRT (Simultaneous Iterative Reconstruction Technique), SART (Simultaneous Algebraic Reconstruction Technique), and CGLS (Conjugate Gradient Least Squares). Users can call the GPU-accelerated projection operations directly from MATLAB or Python to develop custom algorithms. The library is written in C++ and leverages CUDA for GPU acceleration, though CPU-only builds supporting 2D algorithms are available through conda-forge.
Installation is streamlined through conda and pip for Python users on Windows, Linux, and macOS, with binary distributions for Windows and source builds available for Linux and macOS. The project supports Python 3.9-3.13 and MATLAB R2012a or higher, with additional support for Octave. The toolbox is distributed under the GPLv3 license and is maintained by the imec Vision Lab at the University of Antwerp and CWI Amsterdam.