Skip to content

Latest commit

 

History

History
21 lines (15 loc) · 850 Bytes

README.md

File metadata and controls

21 lines (15 loc) · 850 Bytes

Convolutional-Sparse-Coding

(Last update 9.4.2022: stopping criteria)

Run the demo files.

The codes include:

  1. unconstraned CSC algorithm: CSC_unconstrained.m
  2. constraned CSC algorithm: CSC_constrained.m
  3. the consensus ADMM-based CDL method: CDL.m
  4. the consensus ADMM-based multiscale CDL method: CDL_multiscale.m
  5. the ADMM-based CDL method based on direct matrix inversion: CDL_mtx_inv.m
  6. code for generating Gaussian random multiscale dictionaries: initdict.m
  7. code for visualizing multiscale filters: dict2image.m
  8. pre-learned dictionaries (.mat files)

Training images are collected from USC-SIPI database.

Reference : F. G. Veshki and S. A. Vorobyov, "Efficient ADMM-based Algorithms for Convolutional Sparse Coding," in IEEE Signal Processing Letters, doi: 10.1109/LSP.2021.3135196. Email: [email protected]