Paper-PipeDefectSplitting

Code to reproduce results of paper: A Bayesian approach for CT reconstruction with defect detection of subsea pipelines.

Main script

main.py is the main script where the compuations to reproduce the results in the paper are done and results saved.

The main script requires :

1. Installation of CUQIpy. This code runs on a pre-released version of CUQIpy. Install via command: pip install git+https://github.com/CUQI-DTU/CUQIpy.git@sprint16_add_JointModel
2. Installation of Astra (https://github.com/astra-toolbox/astra-toolbox)
3. Real or synthetic data, which is described below.
4. The scripts funs.py and postprocessfuns.py included in this repository.

Plotting

plotting.py creates the plots in the paper.

The plotting script requires:

1. Saved results from numerical experiments in main.py
2. The script plotfuns.py included in this repository.

Data

Synthetic data can be generated using the script generatesynthdata.py Real data is found at https://doi.org/10.5281/zenodo.6817690.

Total variation reconstruction

TV.py contains code to reproduce the TV reconstructions used for comparison in the paper. This script requires installation of CIL (https://tomographicimaging.github.io/CIL/nightly/index.html)

Samples from SGP prior

SGPsample.py contains code to obtain and plot samples from the SGP prior used on the pipe image. Installation of CUQIpy is required. This code runs on a pre-released version of CUQIpy. Install via command: pip install git+https://github.com/CUQI-DTU/CUQIpy.git@sprint16_add_JointModel