Dual-graph convolutional network based on band attention and sparse constraint for hyperspectral band selection

作者：

Highlights：

• An improved graph convolutional network is proposed for band selection.

• Band-based and sample-based graphs are built with spatial and spectral information.

• Design soft-shifting optimization strategy to overcome the optimization problem.

• Mini batches of labeled and unlabeled samples are used for graph construction.

摘要

•An improved graph convolutional network is proposed for band selection.•Band-based and sample-based graphs are built with spatial and spectral information.•Design soft-shifting optimization strategy to overcome the optimization problem.•Mini batches of labeled and unlabeled samples are used for graph construction.

论文关键词：Graph convolutional network,Band selection,Hyperspectral image classification,Attention mechanism

论文评审过程：Received 10 April 2021, Revised 18 August 2021, Accepted 19 August 2021, Available online 24 August 2021, Version of Record 1 September 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.107428