Semi-supervised locality preserving dense graph convolution for hyperspectral image classification

被引：0
|
作者：
Ding Y. [1 ]
Zhang Z. [1 ]
Zhao X. [1 ]
Yang N. [1 ]
Cai W. [2 ]
Cai W. [2 ]
机构：
[1] State Key Scientific Laboratory of Weapon Launch Theory and Technology, Rocket Force University of Engineering, Xi’an
[2] School of Artificial Intelligence and Computer Science, Jiangnan University, Wuxi
来源：
Beijing Hangkong Hangtian Daxue Xuebao/Journal of Beijing University of Aeronautics and Astronautics | 2023年 / 49卷 / 12期
基金：
中国国家自然科学基金;
关键词：
context-aware; graph attention mechanism; graph convolutional network; hyperspectral image classification; superpixel segmentation;
D O I：
10.13700/j.bh.1001-5965.2022.0109
中图分类号：
学科分类号：
摘要：
The application of graph convolutional network (GCN) to hyperspectral image (HSI) classification is the hotspot and frontier of current research. Nevertheless, the over-smoothing, feature adaptive selection, and calculation complexity issues still exist for the graph convolution network approaches that are now accessible. To circumvent these problems, a superpixel segmentation method to reduce the spatial dimension of the HSI is proposal, which reduces the amount of calculation while preserving the spectral characteristics of the nodes. In addition, the dense structure is adopted to retain the features of the convolution in process, and the problem of excessive smoothing of the graph convolution is settled. Finally, a mechanism for extracting the practical local knowledge produced by each layer of the dense GCN is created using a layer-wise context-aware learning approach. The network realizes end-to-end semi-supervised classification. The experimental results on three real datasets show that the proposed algorithm outperforms the compared state-of-the-art methods on all indices and improves the classification accuracy of HSI. © 2023 Beijing University of Aeronautics and Astronautics (BUAA). All rights reserved.
引用
收藏
页码：3409 / 3418
页数：9
相关论文
共 29 条
  • [1] RASTI B, HONG D F, HANG R L, Et al., Feature extraction for hyperspectral imagery: The evolution from shallow to deep, overview and toolbox, IEEE Geoscience and Remote Sensing Magazine, 8, 4, pp. 60-88, (2020)
  • [2] DING Y, ZHAO X F, ZHANG Z L, Et al., Semi-supervised locality preserving dense graph neural network with ARMA filters and context-aware learning for hyperspectral image classification, IEEE Transactions on Geoscience and Remote Sensing, 60, pp. 1-12, (2022)
  • [3] DING Y, ZHAO X F, ZHANG Z L, Et al., Graph sample and aggregate-attention network for hyperspectral image classification, IEEE Geoscience and Remote Sensing Letters, 19, pp. 1-5, (2022)
  • [4] ZHONG P, GONG Z Q, SHAN J X., Multiple instance learning for multiple diverse hyperspectral target characterizations, IEEE Transactions on Neural Networks and Learning Systems, 31, 1, pp. 246-258, (2020)
  • [5] DING Y, ZHAO X F, ZHANG Z L, Et al., Multiscale graph sample and aggregate network with context-aware learning for hyperspectral image classification, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 14, pp. 4561-4572, (2021)
  • [6] BO C J, LU H C, WANG D., Hyperspectral image classification via JCR and SVM models with decision fusion, IEEE Geoscience and Remote Sensing Letters, 13, 2, pp. 177-181, (2016)
  • [7] MA L, CRAWFORD M M, TIAN J W., Local manifold learning-based k-nearest-neighbor for hyperspectral image classification, IEEE Transactions on Geoscience and Remote Sensing, 48, 11, pp. 4099-4109, (2010)
  • [8] DING Y, ZHANG Z L, ZHAO X F, Et al., Deep hybrid: Multigraph neural network collaboration for hyperspectral image classification, Defence Technology, 23, pp. 164-176, (2023)
  • [9] CHEN Y S, LIN Z H, ZHAO X, Et al., Deep learning-based classification of hyperspectral data, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7, 6, pp. 2094-2107, (2014)
  • [10] LI T, ZHANG J P, ZHANG Y., Classification of hyperspectral image based on deep belief networks, 2014 IEEE International Conference on Image Processing, pp. 5132-5136, (2015)
123