Towards Robust Hyperspectral Unmixing: Mixed Noise Modeling and Image-Domain Regularization

This is a demo code of the method proposed in the following reference:

K. Naganuma and S. Ono ``Towards Robust Hyperspectral Unmixing: Mixed Noise Modeling and Image-Domain Regularization''

Update history: Octber 13, 2023: v1.0

For more information, see the following

• Project website: https://www.mdi.c.titech.ac.jp/publications/rhuidr
• Preprint paper: https://arxiv.org/abs/2302.08247

How to use

Run main.m

Contents

Data

We use the HYperspectral Data Retrieval and Analysis (HYDRA) toolbox to generate a synthetic HS image. HYDRA can be obtained at https://www.ehu.eus/ccwintco/index.php?title=Hyperspectral_Imagery_Synthesis_tools_for_MATLAB.

Spectral library

To make an endmember library, we use spectral signatures from the U.S. Geological Survey (USGS) Spectral Library accessed at https://www.usgs.gov/programs/usgs-library.

Reference

If you use this code, please cite the following paper:

@misc{naganuma2023robust,
      title={Towards Robust Hyperspectral Unmixing: Mixed Noise Modeling and Image-Domain Regularization}, 
      author={Kazuki Naganuma and Shunsuke Ono},
      year={2023},
      eprint={2302.08247},
      archivePrefix={arXiv},
      primaryClass={eess.IV}
}