Sketched Multi-view Subspace Learning for Hyperspectral Anomalous Change Detection

This is a demo of this work implemented in Matlab, written by Shizhen Chang, Michael Kopp and Pedram Ghamisi.

For more details, please refer to our paper: Sketched Multi-view Subspace Learning for Hyperspectral Anomalous Change Detection

Environment:

Matlab R2015b

Files

The package contains the following files.

demo.m: A demo shows how to run this work.
SMSL_ACD.m: Implementation of the SMSL model.
jlt.m: Calculate the sketched dictionary through JLT random projection.
meanjlt.m: Calculate the mean sketched dictionary after repeating JLT random projection.
roc_i.m: Calculate the ROC curve.
hyperNormalize.m: Supportive files to normalize the data.

Usage

After unzipping the files, put the current directory of Matlab to mydir.
Run demo.m.

Citation

Please cite our paper if you find it is useful for your research.

@article{chang2022sketched,
  author={Chang, Shizhen and Kopp, Michael and Ghamisi, Pedram},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={Sketched Multi-view Subspace Learning for Hyperspectral Anomalous Change Detection}, 
  year={2022},
  volume={},
  number={},
  pages={1-13},
  doi={10.1109/TGRS.2022.3220814}
  }

Acknowledgment

The authors would like to express their thanks to the creators of Viareggio and BGU-iCVL-hyperspectral-image datasets.

Notice

This repo is distributed under MIT License and is released for scientific purposes only.