Abstract:Sparse unmixing has been successfully applied in hyperspectral remote sensing imagery analysis based on a standard spectral library known in advance. This approach involves reformulating the traditional linear spectral unmixing problem by finding the optimal subset of signatures in this spectral library using the sparse regression technique, and has greatly improved the estimation of fractional abundances in ubiquitous mixed pixels. Since the potentially large standard spectral library can be given a priori, the most challenging task is to compute the regression coefficients, i.e., the fractional abundances, for the linear regression problem. There are many mathematical techniques that can be used to deal with the spectral unmixing problem; e.g., ordinary least squares (OLS), constrained least squares (CLS), orthogonal matching pursuit (OMP), and basis pursuit (BP). However, due to poor prediction accuracy and non-interpretability, the traditional methods often cannot obtain satisfactory estimations or achieve a reasonable interpretation. In this paper, to improve the regression accuracy of sparse unmixing, least angle regression-based constrained sparse unmixing (LARCSU) is introduced to further enhance the precision of sparse unmixing. Differing from the classical greedy algorithms and some of the cautious sparse regression-based approaches, the LARCSU algorithm has two main advantages. Firstly, it introduces an equiangular vector to seek the optimal regression steps based on the simple underlying geometry. Secondly, unlike the alternating direction method of multipliers (ADMM)-based algorithms that introduce one or more multipliers or augmented terms during their optimization procedures, no parameters are required in the computational process of the LARCSU approach. The experimental results obtained with both simulated datasets and real hyperspectral images confirm the effectiveness of LARCSU compared with the current state-of-the-art spectral unmixing algorithms. LARCSU can obtain a better fractional abundance map, as well as a higher unmixing accuracy, with the same order of magnitude of computational effort as the CLS-based methods.

Local Spectral Similarity Preserving Regularized Robust Sparse Hyperspectral Unmixing

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

Hyperspectral Unmixing with Robust Collaborative Sparse Regression

Robust Sparse Hyperspectral Unmixing With Norm

Superpixel-Based Collaborative and Low-Rank Regularization for Sparse Hyperspectral Unmixing

Local Spectral Similarity-Guided Sparse Unmixing of Hyperspectral Images with Spatial Graph Regularization

A NOVEL BAYESIAN LASSO MODEL BASED ON SPATIAL-CORRELATED SPARSITY FOR SEMISUPERVISED HYPERSPECTRAL UNMIXING

Superpixel-based Graph Laplacian Regularized and Weighted Robust Sparse Unmixing

Locally Similar Sparsity-Based Hyperspectral Compressive Sensing Using Unmixing

Robust Multiscale Spectral–Spatial Regularized Sparse Unmixing for Hyperspectral Imagery

Spectral–Spatial Robust Nonnegative Matrix Factorization for Hyperspectral Unmixing

Region-based Collaborative Sparse Unmixing of Hyperspectral Imagery

Half-quadratic Based Robust Hyperspectral Unmixing Framework

An Improved Nonlocal Sparse Unmixing Algorithm for Hyperspectral Imagery

Hyperspectral Unmixing Via Low-Rank Representation With Space Consistency Constraint And Spectral Library Pruning

Non-Local Sparse Unmixing for Hyperspectral Remote Sensing Imagery

Least Angle Regression-Based Constrained Sparse Unmixing Of Hyperspectral Remote Sensing Imagery

Sparsity-Constrained Distributed Unmixing of Hyperspectral Data

Robust Sparse Hyperspectral Unmixing With $\ell_{2,1}$ Norm

Superpixel-Based Reweighted Low-Rank and Total Variation Sparse Unmixing for Hyperspectral Remote Sensing Imagery

Hyperspectral Unmixing Via L-1/2 Sparsity-Constrained Nonnegative Matrix Factorization