Homepage:

https://academic.peterkam.top/

https://liangjiandeng.github.io/

https://sites.google.com/site/vivonegemine/

• Code for paper: "Laplacian pyramid networks: A new approach for multispectral pansharpening, Information Fusion"
• State-of-the-art pansharpening performance

• Python 3.8 (Recommend to use Anaconda)
• TensorFlow 1.14.0
• NVIDIA GPU + CUDA
• Python packages: pip install numpy scipy h5py
• TensorBoard

The datasets used in this paper is WorldView-3 (can be downloaded here), QuickBird (can be downloaded here) and GaoFen-2 (can be downloaded here). Due to the copyright of dataset, we can not upload the datasets, you may download the data and simulate them according to the paper.

Training and testing codes are in 'codes/'. Pretrained model can be found in 'codes/pretrained/'. All codes will be presented after the paper is completed published. Please refer to codes/how-to-run.md for detail description.

FCNN architecture is presented below:

The following quantitative results is generated from WorldView-3 datasets. A.T. is short for Average running Time for saving spaces in the paper.

All quantitative results can be found in 'results/'.

The following visual results is generated from WorldView-3 datasets.

All visual results can be also found in 'results/'.

Part of code of this work is derived from https://xueyangfu.github.io/projects/LPNet.html.

@article{LPPN,
author = {Cheng Jin, Liang-Jian Deng, Ting-Zhu Huang and Gemine Vivone},
title = {Laplacian pyramid networks: A new approach for multispectral pansharpening},
journal = {Information Fusion},
volume = {78},
pages = {158-170},
year = {2022},
issn = {1566-2535},
doi = {https://doi.org/10.1016/j.inffus.2021.09.002}
}

We are glad to hear from you. If you have any questions, please feel free to contact [email protected] or open issues on this repository.

This project is open sourced under GNU Affero General Public License v3.0.