/SCI-TV-FFDNet

Effective Snapshot Compressive-spectral Imaging via Deep Denosing and Total Variation Priors (CVPR 2021)

Primary LanguagePython

SCI Reconstruction with TV and FFDNet

Code for Effective Snapshot Compressive-spectral Imaging via Deep Denoising and Total Variation Priors (CVPR 2021).

Requirements

  • python3
  • pytorch, cvxpy, scipy, numpy, skimage, tqdm

Usage

  • run our algorithm
python pnp_gap_HSI_tv_ffdnet.py
  • run two baseline algorithms
# Run plug-and-play gap based on 3d TV denoiser
python pnp_gap_HSI_3dtv.py
# Run plug-and-play gap based on FFDNet
python pnp_gap_HSI_ffdnet.py