Total Variation Spatial Regularization for Sparse Hyperspectral Unmixing

Marian-Daniel Iordache, José M. Bioucas‐Dias, Antonio Plaza

IEEE Transactions on Geoscience and Remote Sensing · 2012 · 754 citations · 68 references

Concepts

TL;DR

Spectral unmixing estimates fractional abundances of pure spectral signatures in mixed pixels, typically modeled as a sparse linear regression, but prior work has not incorporated spatial information. The study proposes adding total variation regularization to the sparse regression framework to exploit spatial context in hyperspectral images. The authors develop a variable‑splitting augmented‑Lagrangian algorithm that integrates TV regularization into sparse regression for hyperspectral unmixing. Experiments on simulated and real hyperspectral datasets show that incorporating TV improves the characterization of mixed pixels compared to conventional sparse unmixing.

Abstract

Spectral unmixing aims at estimating the fractional abundances of pure spectral signatures (also called endmembers) in each mixed pixel collected by a remote sensing hyperspectral imaging instrument. In recent work, the linear spectral unmixing problem has been approached in semisupervised fashion as a sparse regression one, under the assumption that the observed image signatures can be expressed as linear combinations of pure spectra, known a priori and available in a library. It happens, however, that sparse unmixing focuses on analyzing the hyperspectral data without incorporating spatial information. In this paper, we include the total variation (TV) regularization to the classical sparse regression formulation, thus exploiting the spatial-contextual information present in the hyperspectral images and developing a new algorithm called sparse unmixing via variable splitting augmented Lagrangian and TV. Our experimental results, conducted with both simulated and real hyperspectral data sets, indicate the potential of including spatial information (through the TV term) on sparse unmixing formulations for improved characterization of mixed pixels in hyperspectral imagery.

References

68