Spectral–Spatial Graph Convolutional Networks for Semisupervised Hyperspectral Image Classification

By Anyong Qin, Zhaowei Shang, Jinyu Tian, Yulong Wang, Taiping Zhang, Yuanyan Tang. IEEE Geoscience and Remote Sensing Letters, 2019

333 citations recorded by the source corpus.

Collecting labeled samples is quite costly and time-consuming for hyperspectral image (HSI) classification task. Semisupervised learning framework, which combines the intrinsic information of labeled and unlabeled samples, can alleviate the deficient labeled samples and increase the accuracy of HSI classification. In this letter, we propose a novel semisupervised learning framework that is based on spectral–spatial graph convolutional networks (<inline-formula> <tex-math notation="LaTeX">$\text{S}^{2}$ </tex-math></inline-formula>GCNs). It explicitly utilizes the adjacency nodes in graph to approximate the convolution. In the process of approximate convolution on graph, the proposed method makes full use of the spatial information of the current pixel. The experimental results on three real-life HSI data sets, i.e., Botswana Hyperion, Kennedy Space Center, and Indian Pines, show that the proposed <inline-formula> <tex-math notation="LaTeX">$\text{S}^{2}$ </tex-math></inline-formula>GCN can significantly improve the classification accuracy. For instance, the overall accuracy on Indian data is increased from 66.8% (GCN) to 91.6%.

Subfields: Artificial Intelligence, Media Technology

View Paper: https://www.semanticscholar.org/paper/e9bc01d45ee073df7254b14fecf105a6d0470cc3

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Danfeng Hong
2. Jocelyn Chanussot
3. Jing Yao
4. B. Bak‐Jensen
5. Qichao Liu
6. Ping Zhong
7. Zhihui Wei
8. Chen Gong

In Media Technology

1. Danfeng Hong
2. Jocelyn Chanussot
3. Jing Yao
4. Swalpa Kumar Roy
5. Lefei Zhang
6. Bo Du
7. Xin Wu
8. Jingxiang Yang

Citations in the corpus

• Multiscale Dynamic Graph Convolutional Network for Hyperspectral Image Classification (2019, methodology): “, GCN [34] and spectral–spatial graph convolutional network (S2GCN) [64].”
• MambaHSI: Spatial–Spectral Mamba for Hyperspectral Image Classification (2025, methodology): “Qin et al. [38] simultaneously consider spatial and spectral neighborhoods by extending original GCN to second-order GCN. Wan et al. [39] utilized superpixel segmentation on HS image and fed it into GCN.”
• Hyperspectral Image Classification—Traditional to Deep Models: A Survey for Future Prospects (2021, methodology): “For example, a second-order GCN was proposed in [219] by modeling spatial-spectral relations on manifolds for HSIC by the attempts to reduce the computational cost on graphs.”
• CNN-Enhanced Graph Convolutional Network With Pixel- and Superpixel-Level Feature Fusion for Hyperspectral Image Classification (2020, methodology): “For instance, Qin et al. [41] proposed a semisupervised GCN method, in which the HSI was first encoded into a graph and then processed by a GCN to propagate information between adjacent pixels according to their spectral similarity and spatial distance.”
• A Review of Graph Neural Networks and Their Applications in Power Systems (2021, background): “The key challenges of these spatial-based GCNs are to define convolutional operations with the different number of neighborhoods and to keep the local invariance [25].”