Graph Convolutional Networks for Hyperspectral Image Classification
Institut polytechnique de Grenoble · Centre National de la Recherche Scientifique · +7 more institutions
Abstract
Convolutional neural networks (CNNs) have been attracting increasing attention in hyperspectral (HS) image classification due to their ability to capture spatial-spectral feature representations. Nevertheless, their ability in modeling relations between the samples remains limited. Beyond the limitations of grid sampling, graph convolutional networks (GCNs) have been recently proposed and successfully applied in irregular (or nongrid) data representation and analysis. In this article, we thoroughly investigate CNNs and GCNs (qualitatively and quantitatively) in terms of HS image classification. Due to the construction of the adjacency matrix on all the data, traditional GCNs usually suffer from a huge…
Citation impact
- FWCI
- 201.76
- Percentile
- 100%
- References
- 65
Authors
6- DHDanfeng HongCorresponding
Institut polytechnique de Grenoble, Centre National de la Recherche Scientifique, GIPSA-Lab, Université Grenoble Alpes
- LGLianru Gao
Chinese Academy of Sciences, Aerospace Information Research Institute
- JYJing Yao
Xi'an Jiaotong University
- BZBing Zhang
Chinese Academy of Sciences, Aerospace Information Research Institute, University of Chinese Academy of Sciences
- APAntonio Plaza
Universidad de Extremadura
Topics & keywords
- Computer science
- Hyperspectral imaging
- Bottleneck
- Pattern recognition (psychology)
- Convolutional neural network
- Artificial intelligence
- Adjacency matrix
- Graph