Fusing Hyperspectral and Multispectral Images via Coupled Sparse Tensor Factorization

By Shutao Li, Renwei Dian, Leyuan Fang, José M. Bioucas-Dias. IEEE Transactions on Image Processing, 2018

527 citations recorded by the source corpus.

Fusing a low spatial resolution hyperspectral image (LR-HSI) with a high spatial resolution multispectral image (HR-MSI) to obtain a high spatial resolution hyperspectral image (HR-HSI) has attracted increasing interest in recent years. In this paper, we propose a coupled sparse tensor factorization (CSTF)-based approach for fusing such images. In the proposed CSTF method, we consider an HR-HSI as a 3D tensor and redefine the fusion problem as the estimation of a core tensor and dictionaries of the three modes. The high spatial-spectral correlations in the HR-HSI are modeled by incorporating a regularizer, which promotes sparse core tensors. The estimation of the dictionaries and the core tensor are formulated as a coupled tensor factorization of the LR-HSI and of the HR-MSI. Experiments on two remotely sensed HSIs demonstrate the superiority of the proposed CSTF algorithm over the current state-of-the-art HSI-MSI fusion approaches.

Subfields: Computational Mathematics, Computational Mechanics, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/91eb155a4e8f7eac69cc19851351bcba78c3a1e7

How researchers see this paper

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

In Computational Mathematics

1. Shutao Li
2. Leyuan Fang
3. Renwei Dian

In Computational Mechanics

1. Shutao Li
2. Leyuan Fang
3. Renwei Dian

In Computer Vision and Pattern Recognition

1. Leyuan Fang
2. Renwei Dian
3. Shutao Li
4. Xiao Qian
5. Tai-Xiang Jiang
6. Xile Zhao
7. Yu Liu
8. Xun Chen

Citations in the corpus

• Hyperspectral Image Denoising via Tensor Low-Rank Prior and Unsupervised Deep Spatial–Spectral Prior (2022, background): “observation, urban planning, and agriculture [1], [2], [3], [4] because they have abundant spatial and spectral information.”
• Learning a Low Tensor-Train Rank Representation for Hyperspectral Image Super-Resolution (2019, background): “Li et al. [36] formulate the super-resolution problem as the coupled sparse Tucker decomposition, which alternatively updates the dictionaries of three modes and sparse core tensor.”
• SelfS2: Self-Supervised Transfer Learning for Sentinel-2 Multispectral Image Super-Resolution (2023, background): “However, as pointed out in [11] and [23], the super-resolution of S2 images differs from the conventional pansharpening problem [20], [24], [25], [26], [27], [28] and the hyperspectral image/MSI fusion problem [28], [29], [30], [31] mainly for the high-spatial-resolution bands (i.”