Target-adaptive CNN-based pansharpening

Abstract

We recently proposed a convolutional neural network (CNN) for remote sensing image pansharpening obtaining a significant performance gain over the state of the art. In this paper, we explore a number of architectural and training variations to this baseline, achieving further performance gains with a lightweight network which trains very fast. Leveraging on this latter property, we propose a target-adaptive usage modality which ensures a very good performance also in the presence of a mismatch w.r.t. the training set, and even across different sensors. The proposed method, published online as an off-the-shelf software tool, allows users to perform fast and high-quality CNN-based pansharpening of their own target images on general-purpose hardware.

Reference

G.Scarpa, S.Vitale, D.Cozzolino
Target-adaptive CNN-based pansharpening
IEEE Transactions on Geoscience and Remote Sensing, vol. 56, no. 9, pp. 5443-5457.

ArXiv version: Preprint arXiv: 1709.06054v2


Download 

Python Code on GitHub: https://github.com/sergiovitale/pansharpening-cnn-python-version

[NEW] Matlab (2018b) Code on GitHub: https://github.com/sergiovitale/pansharpening-cnn-matlab-version

BibTex

@ARTICLE{Scarpa2018, 
author={G. Scarpa and S. Vitale and D. Cozzolino}, 
journal={IEEE Transactions on Geoscience and Remote Sensing}, 
title={Target-Adaptive CNN-Based Pansharpening}, 
year={2018}, 
volume={56}, 
number={9}, 
pages={5443-5457}, 
doi={10.1109/TGRS.2018.2817393}, 
ISSN={0196-2892}, 
month={Sept},}