Graph-based Active Learning for Nearly Blind Hyperspectral Unmixing

Bohan Chen, Yifei Lou, Andrea L. Bertozzi, Jocelyn Chanussot

---

The repository implements the method proposed in the paper Graph-based Active Learning for Nearly Blind Hyperspectral Unmixing.

Estimated Abundance Maps for the Jasper Ridge Dataset

Estimated Endmember Spectra for the Jasper Ridge Dataset

Before You Start

System Requirements

• Python 3.8 or 3.9 is required for running this repository.

Install Required Libraries

1. Open your terminal and navigate to the project directory.

2. Run the following command to install the required packages:

   pip install -r requirements.txt

Data Download

If you intend to reproduce the results from the paper or run the examples provided later, you will need to download the necessary dataset. You can download it using the following link: Google Drive Download

How to Run

Precomputed Results

If you just want to get the precomputed results, you can directly run the following command. Please note that you need to download the corresponding data and replace the processed_data folder in this directory before running.

python main.py --dataset jasper

The jasper here can be replaced with samson, urban, and apex.

Run Our Approach on Your Own Dataset

If you would like to run our method on your own dataset, you first need to create a .npy file that contains a dictionary with at least 3 keys: X, A_gt, and S_gt.

• X is an n * m * C HSI image.
• A_gt is the ground truth abundance map of size p * (n * m).
• S_gt is the ground truth endmember spectra of size C * p.

Here, p is the number of endmembers.

After preparing the .npy file, you can then run:

python main.py --dataset path --train_percentage 0.4 --AL_method mcvopt

Here path is the path to the .npy file of your own dataset. train_percentage is the percentage of training pixels, with default value 0.4. AL_method specifies the type of acquisition function to be used in active learning with available choices "vopt", "uc", "mc", "mcvopt".

Note that in our GRSU method, there are some parameters that require tuning. Here, we use a fixed empirical value, but for specific datasets, we recommend using grid search to select the most appropriate parameters.

License

Copyright (C) 2023 Bohan Chen

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see https://www.gnu.org/licenses/.

Contact

If you have any further questions or run into issues, please feel free to reach out to Bohan Chen. You can email him at [email protected].