Mario Parente

Hyperspectral Unmixing Overview: Geometrical, Statistical, and Sparse Regression-Based Approaches

Imaging spectrometers measure electromagnetic energy scattered in their instantaneous field view in hundreds or thousands of spectral channels with higher spectral resolution than multispectral cameras. Imaging spectrometers are therefore often referred to as hyperspectral cameras (HSCs). Higher spectral resolution enables material identification via spectroscopic analysis, which facilitates countless applications that require identifying materials in scenarios unsuitable for classical spectroscopic analysis. Due to low spatial resolution of HSCs, microscopic material mixing, and multiple scattering, spectra measured by HSCs are mixtures of spectra of materials in a scene. Thus, accurate estimation requires unmixing. Pixels are assumed to be mixtures of a few materials, called endmembers. Unmixing involves estimating all or some of: the number of endmembers, their spectral signatures, and their abundances at each pixel. Unmixing is a challenging, ill-posed inverse problem because of model inaccuracies, observation noise, environmental conditions, endmember variability, and data set size. Researchers have devised and investigated many models searching for robust, stable, tractable, and accurate unmixing algorithms. This paper presents an overview of unmixing methods from the time of Keshava and Mustards unmixing tutorial to the present. Mixing models are first discussed. Signal-subspace, geometrical, statistical, sparsity-based, and spatial-contextual unmixing algorithms are described. Mathematical problems and potential solutions are described. Algorithm characteristics are illustrated experimentally.

Phyllosilicate diversity and past aqueous activity revealed at Mawrth Vallis, Mars

Observations by the Mars Reconnaissance Orbiter/Compact Reconnaissance Imaging Spectrometer for Mars in the Mawrth Vallis region show several phyllosilicate species, indicating a wide range of past aqueous activity. Iron/magnesium (Fe/Mg)–smectite is observed in light-toned outcrops that probably formed via aqueous alteration of basalt of the ancient cratered terrain. This unit is overlain by rocks rich in hydrated silica, montmorillonite, and kaolinite that may have formed via subsequent leaching of Fe and Mg through extended aqueous events or a change in aqueous chemistry. A spectral feature attributed to an Fe2+ phase is present in many locations in the Mawrth Vallis region at the transition from Fe/Mg-smectite to aluminum/silicon (Al/Si)–rich units. Fe2+-bearing materials in terrestrial sediments are typically associated with microorganisms or changes in pH or cations and could be explained here by hydrothermal activity. The stratigraphy of Fe/Mg-smectite overlain by a ferrous phase, hydrated silica, and then Al-phyllosilicates implies a complex aqueous history.

A Review of Nonlinear Hyperspectral Unmixing Methods

In hyperspectral unmixing, the prevalent model used is the linear mixing model, and a large variety of techniques based on this model has been proposed to obtain endmembers and their abundances in hyperspectral imagery. However, it has been known for some time that nonlinear spectral mixing effects can be a crucial component in many real-world scenarios, such as planetary remote sensing, intimate mineral mixtures, vegetation canopies, or urban scenes. While several nonlinear mixing models have been proposed decades ago, only recently there has been a proliferation of nonlinear unmixing models and techniques in the signal processing literature. This paper aims to give an historical overview of the majority of nonlinear mixing models and nonlinear unmixing methods, and to explain some of the more popular techniques in detail. The main models and techniques treated are bilinear models, models for intimate mineral mixtures, radiosity-based approaches, ray tracing, neural networks, kernel methods, support vector machine techniques, manifold learning methods, piece-wise linear techniques, and detection methods for nonlinearity. Furthermore, we provide an overview of several recent developments in the nonlinear unmixing literature that do not belong into any of these categories.

Characterization of phyllosilicates observed in the central Mawrth Vallis region, Mars, their potential formational processes, and implications for past climate

Mawrth Vallis contains one of the largest exposures of phyllosilicates on Mars. Nontronite, montmorillonite, kaolinite, and hydrated silica have been identified throughout the region using data from the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM). In addition, saponite has been identified in one observation within a crater. These individual minerals are identified and distinguished by features at 1.38–1.42, ∼1.91, and 2.17–2.41 μm. There are two main phyllosilicate units in the Mawrth Vallis region. The lowermost unit is nontronite bearing, unconformably overlain by an Al-phyllosilicate unit containing montmorillonite plus hydrated silica, with a thin layer of kaolinite plus hydrated silica at the top of the unit. These two units are draped by a spectrally unremarkable capping unit. Smectites generally form in neutral to alkaline environments, while kaolinite and hydrated silica typically form in slightly acidic conditions; thus, the observed phyllosilicates may reflect a change in aqueous chemistry. Spectra retrieved near the boundary between the nontronite and Al-phyllosilicate units exhibit a strong positive slope from 1 to 2 μm, likely from a ferrous component within the rock. This ferrous component indicates either rapid deposition in an oxidizing environment or reducing conditions. Formation of each of the phyllosilicate minerals identified requires liquid water, thus indicating a regional wet period in the Noachian when these units formed. The two main phyllosilicate units may be extensive layers of altered volcanic ash. Other potential formational processes include sediment deposition into a marine or lacustrine basin or pedogenesis.

Mineralogy of the Paso Robles soils on Mars

Abstract Visible, near-infrared, thermal, and Mössbauer spectroscopic data from the exposed, bright track soil at the “Paso Robles” site within Gusev crater, Mars, indicate the presence of Fe3+-sulfates and possibly Fe3+-phosphates admixed with the host soil. When the spectroscopic analyses are combined with constraints imposed by chemical data, the determined dominant Fe3+-sulfate component is hydrous, and all of the spectroscopic methods suggest that it is probably ferricopiapite or some closely related, structurally similar species, possibly mixed with other Fe3+ sulfates such as butlerite or parabutlerite, or perhaps (para)coquimbite, fibroferrite, or metahohmanite. Such an assemblage is consistent with formation in a highly oxidized, relatively dehydrated environment with the bulk-sulfate assemblage having OH/(OH + 2SO4) of < ~0.4. Some Fe3+ is likely to be associated with phosphates in the soil in the form of ferristrunzite or strengite.

Survey of geometric and statistical unmixing algorithms for hyperspectral images

Spectral mixture analysis (also called spectral unmixing) has been an alluring exploitation goal since the earliest days of imaging spectroscopy. No matter the spatial resolution, the spectral signatures collected in natural environments are invariably a mixture of the signatures of the various materials found within the spatial extent of the ground instantaneous field view of the imaging instrument. In this paper, we give a comprehensive enumeration of the unmixing methods used in practice, because of their implementation in widely used software packages, and those published in the literature. We have structured the review according to the basic computational approach followed by the algorithms, with particular attention to those based on the computational geometry formulation, and statistical approaches with a probabilistic foundation. The quantitative assessment of some available techniques in both categories provides an opportunity to review recent advances and to anticipate future developments.

Hyperspectral Image Unmixing via Alternating Projected Subgradients

We consider the problem of factorizing a hyperspectral image into the product of two nonnegative matrices, which represent nonnegative bases for image spectra and mixing coefficients, respectively. This spectral unmixing problem is a nonconvex optimization problem, which is very difficult to solve exactly. We present a simple heuristic for approximately solving this problem based on the idea of alternating projected subgradient descent. Finally, we present the results of applying this method on the 1990 AVIRIS image of Cuprite, Nevada and show that our results are in agreement with similar studies on the same data.

INTERPRETATION OF REFLECTANCE SPECTRA OF CLAY MINERAL-SILICA MIXTURES: IMPLICATIONS FOR MARTIAN CLAY MINERALOGY AT MAWRTH VALLIS

The Al-clay-rich rock units at Mawrth Vallis, Mars, have been identified as mixtures of multiple components based on their spectral reflectance properties and the known spectral character of pure clay minerals. In particular, the spectral characteristics associated with the ~2.2 μm feature in Martian reflectance spectra indicate that mixtures of AlOH- and SiOH-bearing minerals are present. The present study investigated the spectral reflectance properties of the following binary mixtures to aid in the interpretation of remotely acquired reflectance spectra of rocks at Mawrth Vallis: kaolinite-opal-A, kaolinite-montmorillonite, montmorillonite-obsidian, montmorillonite-hydrated silica (opal), and glassillite-smectite (where glass was hydrothermally altered to mixed-layer illite-smectite). The best spectral matches with Martian data from the present study’s laboratory experiments are mixtures of montmorillonite and obsidian having ~50% montmorillonite or mixtures of kaolinite and montmorillonite with ~30% kaolinite. For both of these mixtures the maximum inflection point on the long wavelength side of the 2.21 μm absorption feature is shifted to longer wavelengths, and in the case of the kaolinite-montmorillonite mixtures the 2.17 μm absorption found in kaolinite is of similar relative magnitude to that feature as observed in CRISM (Compact Reconnaissance Imaging Spectrometer for Mars) data. The reflectance spectra of clay mixed with opal and of hydrothermally altered glass-illite-smectite did not represent the Martian spectra observed in this region as well. A spectral comparison of linear vs. intimate mixtures of kaolinite and montmorillonite indicated that for these sieved samples, the intimate mixtures are very similar to the linear mixtures with the exception of the altered glass-illite-smectite samples. However, the 2.17 μm kaolinite absorption is stronger in the intimate mixtures than in the equivalent linear mixture. Modified Gaussian Modeling of absorption features observed in reflectance spectra of the kaolinite-montmorillonite mixtures indicated a strong correlation between percent kaolinite in the mixture and the ratio of the area of the 2.16 μm band found in kaolinite to the area of the 2.20 μm band found in montmorillonite.

The mawrth vallis region of mars: A potential landing site for the mars science laboratory (MSL) mission

The primary objective of NASAs Mars Science Laboratory (MSL) mission, which will launch in 2011, is to characterize the habitability of a site on Mars through detailed analyses of the composition and geological context of surface materials. Within the framework of established mission goals, we have evaluated the value of a possible landing site in the Mawrth Vallis region of Mars that is targeted directly on some of the most geologically and astrobiologically enticing materials in the Solar System. The area around Mawrth Vallis contains a vast (>1 × 10⁶ km²) deposit of phyllosilicate-rich, ancient, layered rocks. A thick (>150 m) stratigraphic section that exhibits spectral evidence for nontronite, montmorillonite, amorphous silica, kaolinite, saponite, other smectite clay minerals, ferrous mica, and sulfate minerals indicates a rich geological history that may have included multiple aqueous environments. Because phyllosilicates are strong indicators of ancient aqueous activity, and the preservation potential of biosignatures within sedimentary clay deposits is high, martian phyllosilicate deposits are desirable astrobiological targets. The proposed MSL landing site at Mawrth Vallis is located directly on the largest and most phyllosilicate-rich deposit on Mars and is therefore an excellent place to explore for evidence of life or habitability.

End-to-End Simulation and Analytical Model of Remote-Sensing Systems: Application to CRISM

The simulation of remote-sensing hyperspectral images is a useful tool for a variety of tasks such as the design of systems, the understanding of the image formation process, and the development and validation of data processing algorithms. The lack of ground truth and the incomplete knowledge of the Martian environment make simulation studies of Mars hyperspectral images a useful tool for automated analysis of Mars data. Hyperspectral near-infrared scenes of mineral mixtures have been simulated to analyze the contributions of surface minerals, atmosphere, and sensor noise on images of Mars. Modeling the remote-sensing process creates a means for the independent analysis of the influence of the environment and instruments on the detection accuracy of the surface composition (e.g., the scene endmembers). The end-to-end model builds surface reflectance scenes based on laboratory sample spectra, creates atmospheric effects using radiative transfer routines, simulates the instrument response function using CRISM data files, and adds instrument noise from thermal and other sources. The purpose of this paper is to understand the hyperspectral remote-sensing process to eventually enable the elevated detection accuracy of minerals on the surface of Mars. The viability of a linear approximation of the complete model is also investigated. The approximation is compared to the complete model in an image classification task.