/Quaternion-Sparse-Coding

Sparse coding and dictionary learning based on quaternion algebra

Primary LanguageMatlabGNU General Public License v3.0GPL-3.0

Quaternion-Sparse-Coding

Sparse coding and dictionary learning based on quaternion algebra

Short descriptions

Quaternion dictionary training in KQSVD:

  • The main functions include main_DictTraining.m and main_Reconstruction.m. main_DictTraining.m: Calling function for dictionary training step in K-QSVD algorithm. The trained disctionary is generated in folder of "training result". main_Reconstruction.m: Calling function for color image reconstruction using K-QSVD algorithm.
  • The other called functions: Type "help XXX" will show the help information of this function.
  • Subfolder of Dataset classifying: For Storage of the input training image and reconstruction image. Here the training images are “flower” while the reconstructed images are “animal”, which verifies the robustness of our method to the variety of patches’ source.
  • Subfolder of Training result: for Storage of the trained dictionary.

Color Image denoising using KQSVD:

  • main_Denoising.m: Calling function for Color Image De-noising using K-QSVD.
  • The other called functions: Type "help XXX" will show the help information of this function.
  • Subfolder of original: For Storage of the input image.
  • Subfolder of Training result: For Storage of the trained dictionary without noise.

Reference and citations

Please cite following papers for utilizing the codes: