Abstract:Hyperspectral (HS) unmixing is the process of decomposing an HS image into material-specific spectra (endmembers) and their spatial distributions (abundance maps). Existing unmixing methods have two limitations with respect to noise robustness. First, if the input HS image is highly noisy, even if the balance between sparse and piecewise-smooth regularizations for abundance maps is carefully adjusted, noise may remain in the estimated abundance maps or undesirable artifacts may appear. Second, existing methods do not explicitly account for the effects of stripe noise, which is common in HS measurements, in their formulations, resulting in significant degradation of unmixing performance when such noise is present in the input HS image. To overcome these limitations, we propose a new robust hyperspectral unmixing method based on constrained convex optimization. Our method employs, in addition to the two regularizations for the abundance maps, regularizations for the HS image reconstructed by mixing the estimated abundance maps and endmembers. This strategy makes the unmixing process much more robust in highly-noisy scenarios, under the assumption that the abundance maps used to reconstruct the HS image with desirable spatio-spectral structure are also expected to have desirable properties. Furthermore, our method is designed to accommodate a wider variety of noise including stripe noise. To solve the formulated optimization problem, we develop an efficient algorithm based on a preconditioned primal-dual splitting method, which can automatically determine appropriate stepsizes based on the problem structure. Experiments on synthetic and real HS images demonstrate the advantages of our method over existing methods.

What problem does this paper attempt to address?

The problems that this paper attempts to solve mainly focus on two key challenges in Hyperspectral Unmixing (HSU): 1. **Degraded unmixing performance under high noise levels**: When the input hyperspectral image is contaminated by high - level noise, even if the balance between sparse regularization and piecewise - smooth regularization is carefully adjusted, a large amount of noise may still remain in the estimated abundance map or unwanted artifacts may appear. Existing unmixing methods perform poorly in highly noisy scenarios. 2. **Inability to effectively handle stripe noise**: Existing methods do not explicitly consider the influence of stripe noise in their formulations, and this type of noise is very common in hyperspectral measurements. Therefore, when stripe noise exists in the input hyperspectral image, the unmixing performance will be significantly degraded. To overcome these limitations, the paper proposes a new robust hyperspectral unmixing method based on constrained convex optimization (Robust Hyperspectral Unmixing using Image - Domain Regularization, RHUIDR). This method not only uses sparse and piecewise - smooth regularization of the abundance map but also introduces a strategy for regularizing the hyperspectral image reconstructed by the abundance map estimated through mixing and endmembers. This strategy makes the unmixing process more robust in highly noisy scenarios and can handle a wider range of noise types, including stripe noise. Specifically, the main contributions of the paper include: 1. **Robustness to high - level noise**: By introducing image - domain regularization, RHUIDR improves the robustness of unmixing in highly noisy scenarios. 2. **Robustness to mixed noise (including stripe noise)**: Explicitly modeling three types of noise (Gaussian noise, sparse noise, and stripe noise) enables RHUIDR to effectively handle mixed noise. 3. **Easy adjustment of hyper - parameters**: By modeling the data fidelity and noise terms as hard constraints instead of adding them to the objective function, the constrained form of RHUIDR decouples the interdependent hyper - parameters, thereby simplifying parameter settings. 4. **Avoiding manual adjustment of step size**: Different from the optimization algorithms used in existing unmixing methods, the algorithm of RHUIDR based on the Preconditioned Primal - Dual Splitting (P - PDS) can automatically determine an appropriate step size according to the problem structure. Experimental results show that RHUIDR outperforms existing methods on synthetic and real hyperspectral images.

Towards Robust Hyperspectral Unmixing: Mixed Noise Modeling and Image-Domain Regularization

Spectral–Spatial Robust Nonnegative Matrix Factorization for Hyperspectral Unmixing

Robust Hyperspectral Unmixing with Practical Learning-Based Hyperspectral Image Denoising

Correntropy-Based Spatial-Spectral Robust Sparsity-Regularized Hyperspectral Unmixing

Robust Nonlinear Unmixing for Hyperspectral Images Based on an Extended Multilinear Mixing Model

Robust Nonlinear Unmixing Of Hyperspectral Images With A Linear-Mixture/Nonlinear-Fluctuation Model

Two-Dimensional Robust Nonnegative Matrix Factorization for Hyperspectral Unmixing.

Half-quadratic Based Robust Hyperspectral Unmixing Framework

Hyperspectral Unmixing Network Accounting for Spectral Variability Based on a Modified Scaled and a Perturbed Linear Mixing Model

Spectral–Spatial Reweighted Robust Nonlinear Unmixing for Hyperspectral Images Based on an Extended Multilinear Mixing Model

Noise Modeling and Learning-Based Hyperspectral Image Denoising Used for Hyperspectral Unmixing

Hyperspectral Unmixing via Noise-Free Model

Robust Multiscale Spectral–Spatial Regularized Sparse Unmixing for Hyperspectral Imagery

Hypergraph-Regularized Sparse NMF for Hyperspectral Unmixing.

Spectral and Spatial Complexity-Based Hyperspectral Unmixing

Hyperspectral Unmixing Using Robust Deep Nonnegative Matrix Factorization

Reweighted Sparse Unmixing for Hyperspectral Images with Noise Level Estimation.

Kernel Based Sparse NMF Algorithm for Hyperspectral Unmixing

Hyperspectral Unmixing with Scaled and Perturbed Linear Mixing Model to Address Spectral Variability

A Robust Multilinear Mixing Model with <i>l</i><sub>2,1</sub> norm for Unmixing Hyperspectral Images

Robust Linear Unmixing for Hyperspectral Remote Sensing Imagery Based on Enhanced Constraint of Classification