qichaoliu / MDGCN

Multiscale Dynamic Graph Convolutional Network for hyperspectral image classification

MDGCN

Description

This is the repository for the TGRS paper [Multiscale Dynamic Graph Convolutional Network for Hyperspectral Image Classification].

Requirements

• Tensorflow (1.4.0)

Usage

You can conduct classification experiments on hyperspectral images (e.g., Indian Pines) by running the 'Main.m' file.

Cite

Please cite our paper if you use this code in your own work:

@ARTICLE{8907873, 
    author={S. {Wan} and C. {Gong} and P. {Zhong} and B. {Du} and L. {Zhang} and J. {Yang}}, 
    journal={IEEE Transactions on Geoscience and Remote Sensing}, 
    title={Multiscale Dynamic Graph Convolutional Network for Hyperspectral Image Classification}, 
    year={2019}, 
    volume={}, 
    number={}, 
    pages={1-16}, 
    keywords={Hyperspectral imaging;Convolution;Feature extraction;Kernel;Support vector machines;Training;Dynamic graph;graph convolutional network (GCN);hyperspectral image classification;multiscale information.}, 
    doi={10.1109/TGRS.2019.2949180}, 
    ISSN={1558-0644}, 
    month={}
}