Hierarchical Superpixel Relation Graph Combined with Convolutional Sparse Coding for Self-Supervised Hyperspectral Image Denoising
- Resource Type
- Conference
- Authors
- Jiang, Zhongshun; Qian, Qipeng; Qiu, Yi; Qian, Yuntao
- Source
- IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium Geoscience and Remote Sensing Symposium, IGARSS 2023 - 2023 IEEE International. :5379-5382 Jul, 2023
- Subject
- Aerospace
Components, Circuits, Devices and Systems
Fields, Waves and Electromagnetics
Geoscience
Signal Processing and Analysis
Convolutional codes
Image coding
Noise reduction
Supervised learning
Self-supervised learning
Feature extraction
Graph neural networks
Hyperspectral images denoising
self-supervised learning
convolutional sparse coding
graph neural network
- Language
- ISSN
- 2153-7003
Self-supervised methods have recently been widely used for hyperspectral image (HSI) denoising. As only a single noisy HSI required to be restored is used for learning, effectively exploring the spatial-spectral joint information for HSI representation becomes to be a more critical problem. In this paper, we propose a self-supervised denoiser based on a blind-spot network and a new feature extraction method that combines convolutional neural network (CNN) and graph neural network (GNN). The features from the convolutional backbone network are fed into the GNN as graph nodes. With hierarchical superpixel segmentation of HSI, the feature nodes with contextual and hierarchical relationships are feature-enhanced by the powerful feature interaction capability of GNN. Experimental results show that our method is competitive compared with state-of-the-art methods.