Dynamical Hyperspectral Unmixing with Variational Recurrent Neural Networks

This package contains the implementation of ReDSUNN algorithm proposed in the paper [1].

Multitemporal hyperspectral unmixing (MTHU) reveals the dynamical evolution of the materials (endmembers) and of their proportions (abundances) in a scene. In this work, we propose an unsupervised MTHU algorithm based on variational recurrent neural networks accounting for the spatial and temporal variability of the endmembers using a low-dimensional parametrization. An MTHU algorithm based on deep variational inference is proposed to tackle the nonlinearity of the model, in which the posterior distribution of the estimated abundances and endmembers is represented using recurrent neural networks and a physically motivated model.

The code is implemented in Pytorch, and was adapted from this repository. The evaluation of the result via quantitative metrics and visualization is performed via MATLAB.

Usage

To reproduce the results of the paper, first run train.py with the appropriate json configuration file, for example:

python train.py -c config_synth_ex1.json

The results are saved in .mat format in the saved folder. Then, you can evaluate the saved results with the corresponding Matlab function evaluate_results_ex1.m.

Downloading the datasets

The datasets are available in Zenodo here, https://doi.org/10.5281/zenodo.7796597. Please download them and include the correct paths to them in the .json files to be able to run the examples.

Running with other data

To run the method with a different dataset, you need:

• .mat files with the hyperspectral image, and the reference/average endmember initialization (using VCA).
• a .json configuration file specifying the parameters of the method and the path to the data files (see examples config_synth_ex1.json, config_synth_ex2.json and config_Tahoe.json).
• a data loader files, stored in the data_loader folder. First, a data loader file has to be created according to the examples in mydataloader_synth_ex1.py, mydataloader_synth_ex2.py and mydataloader_Tahoe.py. Then, a corresponding dataloader class must be added to the file data_loaders.py. Note that the path to the .mat file containing the image is passed to the data loaders from the .json file.
• a Matlab file to evaluate the resulta, if desired, according to the examples evaluate_results_ex1.m, evaluate_results_ex2.m, evaluate_results_ex_Tahoe.m.

REFERENCE:

If you use this software please cite the following in any resulting publication:

[1] Dynamical Hyperspectral Unmixing with Variational Recurrent Neural Networks
    R.A. Borsoi, T. Imbiriba, P Closas.
    IEEE Transactions on Image Processing, 2023.