Total Deep Variation Regularizer

Here, we showcase the total deep variation (TDV) regularizer introduced in

• E. Kobler, A. Effland, K. Kunisch, and T. Pock, Total deep variation for linear inverse problems, in IEEE Conference on Computer Vision and Pattern Recognition, 2020.
• E. Kobler, A. Effland, K. Kunisch, and T. Pock, Total Deep Variation: A Stable Regularizer for Inverse Problems, arXiv preprint arXiv:2006.08789.

If you use this code please cite:

@InProceedings{KoEf20,
  Title     = {Total Deep Variation for Linear Inverse Problems},
  Author    = {Kobler, Erich and Effland, Alexander and Kunisch, Karl and Pock, Thomas},
  Booktitle = {IEEE Conference on Computer Vision and Pattern Recognition},
  Year      = {2020}
}

Overview

.
+-- ddr               : data driven regularizers module
| +-- conv.py         : implementation of forward/backward convolution operators
| +-- regularizer.py  : interface for regularizers
| +-- tdv.py          : implementation of the TDV regularizer
+-- data              : sample images
+-- checkpoints       : pytorch checkpoint files
+-- figures           : figures for plotting
+-- denoise.py        : simple script to run gray-scale/color denoising
+-- eigenfunctions.py : visualization of an eigenfunction of the TDV regularizer

Requirements

• numpy
• imageio
• pytorch
• scikit-image
• optox
• matplotlib

Gaussian Image Denoising

Single Image Super-resolution

Non-linear Eigenfunction Analysis