Fusing Hyperspectral and Multispectral Images via Coupled Sparse Tensor Factorization.

Area of Science:

• Remote Sensing
• Image Processing
• Data Fusion

Background:

• Hyperspectral image (HSI) and multispectral image (MSI) fusion is crucial for enhancing spatial and spectral resolutions.
• Existing fusion methods often struggle to fully exploit the complex spatial-spectral correlations within hyperspectral data.

Purpose of the Study:

• To develop a novel approach for fusing low spatial resolution hyperspectral images (LR-HSI) with high spatial resolution multispectral images (HR-MSI).
• To achieve a high spatial resolution hyperspectral image (HR-HSI) by effectively modeling spatial-spectral correlations.

Main Methods:

• Proposed a coupled sparse tensor factorization (CSTF) method, treating the HR-HSI as a three-dimensional tensor.
• Redefined the fusion problem as estimating a sparse core tensor and dictionaries for the three modes.
• Employed a regularizer to promote sparse core tensors, capturing high spatial-spectral correlations, and coupled tensor factorization for dictionary and core tensor estimation.

Main Results:

• The CSTF algorithm demonstrated superior performance compared to existing state-of-the-art HSI-MSI fusion methods.
• Experiments on two remotely sensed hyperspectral datasets validated the effectiveness of the proposed fusion approach.

Conclusions:

• The proposed CSTF method offers an effective solution for HSI-MSI fusion, yielding high-resolution hyperspectral images.
• The approach successfully models and leverages spatial-spectral correlations for improved fusion outcomes.