Image Reconstruction Source Code

What does this package contain

Beam Hardening Correction Algorithms

All examples and data for our blind beam hardening correction method are under folder bhcEx with filenames ended with Ex. The figures in our paper can be reproduced by first run *Ex, e.g., yangEx, followed by *Ex('plot'), e.g., yangEx('plot'). Algorithm implementations are under bhc.

References

1. R. Gu and A. Dogandžić, “Blind X-ray CT Image Reconstruction from Polychromatic Poisson Measurements,” IEEE Trans. Comput. Imag., vol. 2, no. 2, pp. 150–165, 2016. [DOI] [PDF] [Poster] [Presentation Video]

2. R. Gu and A. Dogandžić. (Sep. 2015). Polychromatic X-ray CT Image Reconstruction and Mass-Attenuation Spectrum Estimation. arXiv: 1509.02193.

3. R. Gu and A. Dogandžić, “Beam hardening correction via mass attenuation discretization,” in Proc. IEEE Int. Conf. Acoust., Speech, Signal Process., Vancouver, Canada, May 2013, pp. 1085–1089. [DOI] [PDF] [Poster]

Projected Nesterov's Proximal-Gradient (PNPG) algorithm

Although being used as a tool to solve the beam hardening problem, PNPG has evolved and now is living in a separate repository.

How to Install

To install this package, first download the repository by running

git clone https://github.com/isucsp/imgRecSrc.git

after downloading, from MATLAB change your current folder to imgRecSrc/ and execute setupPath.m to add necessary paths to the environment.

For X-ray CT examples, the projection and back projection operator subroutines may be called from MATLAB. Since they are written in c language, to prepare MATLAB recognizable MEX files, go to imgRecSrc/prj and compile the necessary files. Instructions on compiling the code are provided for both UNIX and Windows:

For UNIX

require: gcc, cuda toolkit (optional) and GPU (optional)

Execute make cpu to compile all cpu implementations. If you have GPU equipped, run make gpu to compile GPU implementation of the X-ray CT projection operators. The matlab code will automatically choose to run on GPU if equipped.

If errors are reported while compiling the *.c/*.cu files under imgRecSrc/prj, please edit the first few lines in imgRecSrc/prj/Makefile to make sure the path for your CUDA installation is correct.

Don't forget to add the path to CUDA library before starting your Matlab:

export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

which can be added to your .bashrc file without harm.

For Windows

require: Visual Studio, cuda toolkit (optional) and GPU (optional)

Execute imgRecSrc/setupPath.m can automatically compile all needed files and add paths.

If there is a GPU equipped in your PC, follow the following steps:

• Open the VS Native Tools Command Prompt via Start -> Microsoft Visual Studio -> Visual Studio Tools;

• Use cd command to change directory to your imgRecSrc/prj;

• Run nvcc -c gpuPrj.cu to generate the obj file;

• Execute imgRecSrc/setupPath.m.

Note

The comments in some of *.m files may contain greek letters, which are UTF-8 encoded. Please open in an appropriately configured text editor.