Search NASA⌕ Search

SEARCH · Search NASA

Results for “Spectral feature extraction”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 199 records · Page 11

Machine learning of Kondo physics using variational autoencoders and symbolic regression

In this work, we employ variational autoencoders to extract physical insight from a dataset of one-particle Anderson impurity model spectral functions. Autoencoders are trained to find a low-dimensional, latent space representation that faithfully characterizes each element of the training set, as measured by a reconstruction error. Variational autoencoders, a probabilistic generalization of standard autoencoders, further condition the learned latent space to promote highly interpretable features. In our study, we find that the learned latent variables strongly correlate with well known, but nontrivial, parameters that characterize emergent behaviors in the Anderson impurity model. In particular, one latent variable correlates with particle-hole asymmetry, while another is in near one-to-one correspondence with the Kondo temperature, a dynamically generated low-energy scale in the impurity model. Using symbolic regression, we model this variable as a function of the known bare physical input parameters and "rediscover" the nonperturbative formula for the Kondo temperature. The machine learning pipeline we develop suggests a general purpose approach, which opens opportunities to discover new domain knowledge in other physical systems.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

IRAS low resolution spectra of 26 symbiotic stars

Data related to the spectral scans for 26 symbiotic stars are described which were extracted from the IRAS low resolution database. Data from the 8-15- and 15-23-micron bands are merged in a program that scales the longer wavelength and produces a weighted average of the spectral scans for each source. The survey shows that active dust producers can probably be isolated and some theories related to the presence of dust emission features are discussed in terms of source variability for measurements made with low resolution spectra.

Stencel, Robert E.↗

Vegetation Biochemistry: What Can Imaging Spectrometry Tell Us About Canopies?

Changes in ecosystem processes such as productivity and decomposition may be expressed in the canopy foliar chemistry resulting from altered carbon allocation patterns, metabolic processes and nutrient availability. Understanding carbon balance on land over large regions requires quantitative determination of leaf constituents such as lignin and total nitrogen from remote sensing imaging systems. Results from spectral reflectance measurements of stacked leaves in the laboratory show that spectrum matching techniques are applicable to the derivation of the equivalent liquid water thickness in plants as well as to the extraction of dry leaf matter reflectance spectra from spectra of green leaves. The residual spectra derived by subtracting water spectra from the spectra of green leaves shows a feature at 1.72 micrometers that can be related to the lignin content of the leaves. Oak leaves have a deeper residual absorption feature than do cotton leaves which is consistent with their relative lignin content. Similar results are achieved when deriving the residuals from images taken over areas of grass and pine trees. Imaging spectrometry provides promise in developing images of various foliar biochemical constituents.

Goetz, Alexander F. H.↗

Spectral Monitoring of Volatiles During Drilling into Frozen Lunar Simulant

NASA's Resource Prospector (RP) project intends to characterize the 3D distribution of volatiles in permanently shadowed regions at the lunar poles. One RP remote sensing instrument is a near-infrared spectrometer with an associated camera and radiometer, called the Near-InfraRed Volatile Spectrometer System (NIRVSS). In May 2016, NIRVSS, a Honeybee Robotics drill, and an Inficon mass spectrometer were placed in a vacuum chamber at Glenn Research Center. Also inside was a tube (1.2 m high x 25 cm diameter) filled with lunar simulant NU-LHT-3M, initially doped with a homogeneous water abundance of ~5%, chilled to cryogenic temperatures and exposed to a vacuum (~10e-6 Torr). During drilling, the NIRVSS instruments observed the cuttings pile as subsurface materials were emplaced on the surface. Spectral features associated with water ice, near 2000 and 3000 nm, were measured by the spectrometer during drilling. The spectral data documents development of a desiccated soil layer in the tube down to ~25-30 cm (confirmed by post-test soil analyses), formed during the initial pump down to vacuum. Drilling occurred in 10 cm segments, with the drill stem extracted and flutes brushed after each 10 cm depth. One exception to this was the 40 cm depth segment where the soil was delivered to a sample capture mechanism, and sealed for post-test analyses. To ~30 cm depth the greatest 2000 and 3000 nm signatures were associated with brushing of the drill flutes above the surface. At depths >40 cm the strongest ice signatures were associated with the drill clearing soil from the existing hole, or beginning to encounter new material. For these greater depths, brushing the flutes after extraction produced much weaker ice signatures than for shallower depths. This suggests that the soil may remain trapped in the exit funnel and is not emplaced on the surface. After each event creating strong ice signatures, these signatures decreased to near background levels in 5 minutes or less, due to surface exposure to vacuum.

lunar poles↗

Development of a quantitative basis for selection of spectral features in a vegetation monitoring system

The development of an objective methodology for evaluation of alternative Landsat data preprocessing options, spectral transform features for monitoring vegetation, and feature summarization algorithms is presented. Based on estimates of spectral separability between a target class and its confusion classes, analysis of variance techniques are used to evaluate potential design options for large scale vegetation monitoring systems. Case studies are presented for early season and through the season spring small grains separation and for barley/other spring small grains separation. It is concluded that a basis for efficient, objective selection among alternative feature extraction approaches has been established for the large scale vegetation mapping/inventory problem. Although the approach has been demonstrated for the unitemporal class separability case, extensions to the multitemporal case are under development.

Phinney, D. E.↗

Feature detection in satellite images using neural network technology

A feasibility study of automated classification of satellite images is described. Satellite images were characterized by the textures they contain. In particular, the detection of cloud textures was investigated. The method of second-order gray level statistics, using co-occurrence matrices, was applied to extract feature vectors from image segments. Neural network technology was employed to classify these feature vectors. The cascade-correlation architecture was successfully used as a classifier. The use of a Kohonen network was also investigated but this architecture could not reliably classify the feature vectors due to the complicated structure of the classification problem. The best results were obtained when data from different spectral bands were fused.

Augusteijn, Marijke F.↗

Deep learning-driven super-resolution in Raman hyperspectral imaging: Efficient high-resolution reconstruction from low-resolution data

Deep learning (DL) has become an indispensable tool in hyperspectral data analysis, automatically extracting valuable features from complex, high-dimensional datasets. Super-resolution reconstruction, an essential aspect of hyperspectral data, involves enhancing spatial resolution, particularly relevant to low-resolution hyperspectral data. Yet, the pursuit of super-resolution in hyperspectral analysis is fraught with challenges, including acquiring ground truth high-resolution data for training, generalization, and scalability. The pressing issue of extended spectral acquisition times, notably for high-resolution scans, is a significant roadblock in hyperspectral imaging. Super-resolution methods offer a promising solution by providing higher spatial resolution data to expedite data collection and yield more efficient outcomes. This paper delves into a practical application of these concepts using Raman imaging, where spectral acquisition times can be prohibitively long. In this context, DL-based super-resolution models demonstrate their efficacy by predicting and reconstructing high-resolution Raman data from low-resolution input, eliminating the need for resource-intensive high-resolution scans. While previous work often relied on substantial high-resolution datasets, this study showcases the ability to achieve similar outcomes even with limited data, presenting a more practical and cost-effective approach. In conclusion, the results offer a glimpse into the transformative potential of this technology to streamline hyperspectral imaging applications by saving valuable time and resources through the successful generation of high-resolution data from low-resolution inputs.

42 ENGINEERING↗

A Shifted Block Lanczos Algorithm 1: The Block Recurrence

In this paper we describe a block Lanczos algorithm that is used as the key building block of a software package for the extraction of eigenvalues and eigenvectors of large sparse symmetric generalized eigenproblems. The software package comprises: a version of the block Lanczos algorithm specialized for spectrally transformed eigenproblems; an adaptive strategy for choosing shifts, and efficient codes for factoring large sparse symmetric indefinite matrices. This paper describes the algorithmic details of our block Lanczos recurrence. This uses a novel combination of block generalizations of several features that have only been investigated independently in the past. In particular new forms of partial reorthogonalization, selective reorthogonalization and local reorthogonalization are used, as is a new algorithm for obtaining the M-orthogonal factorization of a matrix. The heuristic shifting strategy, the integration with sparse linear equation solvers and numerical experience with the code are described in a companion paper.

Grimes, Roger G.↗

Wavelet analysis of molecular dynamics: Efficient extraction of time-frequency information in ultrafast optical processes

New experimental techniques based on nonlinear ultrafast spectroscopies have been developed over the last few years, and have been demonstrated to provide powerful probes of quantum dynamics in different types of molecular aggregates, including both natural and artificial light harvesting complexes. Fourier transform-based spectroscopies have been particularly successful, yet “complete” spectral information normally necessitates the loss of all information on the temporal sequence of events in a signal. This information though is particularly important in transient or multi-stage processes, in which the spectral decomposition of the data evolves in time. By going through several examples of ultrafast quantum dynamics, we demonstrate that the use of wavelets provide an efficient and accurate way to simultaneously acquire both temporal and frequency information about a signal, and argue that this greatly aids the elucidation and interpretation of physical process responsible for non-stationary spectroscopic features, such as those encountered in coherent excitonic energy transport.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Dust around Mira variables: An analysis of IRAS LRS spectra

The spatial extent and spectral appearance of the thin dust shell around Mira variables is determined largely by the dust absorptivity, Q(sub abs)(lambda), and the dust condensation temperature T(sub cond). Both Q(sub abs)(lambda) and T(sub cond) are extracted from IRAS low-resolution spectra (LRS) spectra. In order to do this, the assumption that the ratio of total power in the 10 micron feature to that in the 20 micron feature should be equal to that measured in other amorphous silicates (e.g., synthesized amorphous Mg2SiO4). It was found that T(sub cond) decreases with decreasing strength of the 10 micron feature, from T(sub cond) = 1000 K to 500 K (estimated error 20 percent). A value for the near-infrared dust absorptivity could not be determined. Although this parameter strongly affects the condensation radius, it hardly affects the shape of the LRS spectrum (as long as the optically thin approximation is valid), because it scales the spatial distribution of the dust. Information on the magnitude of the near-infrared dust absorptivity may be deduced from the unique carbon star BM Gem. This star has a LRS spectrum with silicate features indication an inner dust shell temperature of at least 1000 K. However, on the basis of observations in the 1920s-30s one may infer an inner dust shell radius of at least 6x10(exp 12)m. To have this high temperature at such a large distance, the near-infrared absorptivity of the dust must be high.

Slijkhuis, S.↗

Extracting scene feature vectors through modeling, volume 3

The remote estimation of the leaf area index of winter wheat at Finney County, Kansas was studied. The procedure developed consists of three activities: (1) field measurements; (2) model simulations; and (3) response classifications. The first activity is designed to identify model input parameters and develop a model evaluation data set. A stochastic plant canopy reflectance model is employed to simulate reflectance in the LANDSAT bands as a function of leaf area index for two phenological stages. An atmospheric model is used to translate these surface reflectances into simulated satellite radiance. A divergence classifier determines the relative similarity between model derived spectral responses and those of areas with unknown leaf area index. The unknown areas are assigned the index associated with the closest model response. This research demonstrated that the SRVC canopy reflectance model is appropriate for wheat scenes and that broad categories of leaf area index can be inferred from the procedure developed.

Berry, J. K.↗

Spatial variation analyses of Thematic Mapper data for the identification of linear features in agricultural landscapes

A need exists for digitized information pertaining to linear features such as roads, streams, water bodies and agricultural field boundaries as component parts of a data base. For many areas where this data may not yet exist or is in need of updating, these features may be extracted from remotely sensed digital data. This paper examines two approaches for identifying linear features, one utilizing raw data and the other classified data. Each approach uses a series of data enhancement procedures including derivation of standard deviation values, principal component analysis and filtering procedures using a high-pass window matrix. Just as certain bands better classify different land covers, so too do these bands exhibit high spectral contrast by which boundaries between land covers can be delineated. A few applications for this kind of data are briefly discussed, including its potential in a Universal Soil Loss Equation Model.

Pelletier, R. E.↗

CHANDRA Detection of the AM CVn Binary ES Cet (KUV 01584-0939)

We report on Chandra ACE observations of the ultracompact AM CVn binary ES Cet. This object has a 10.3 minute binary period and is the most compact of the confirmed AM CVn systems. We have, for the first time, unambiguously detected the X-ray counterpart to ES Cet. In a 20 ksec ACIS-S image a point-like X-ray source is found within 1 sec. of the catalogued optical position. The mean countrate in ACIS-S is 0.013/s, and there is no strong evidence for variability. We folded the X-ray data using the optical ephemeris of Warner & Woudt, but did not detect any significant modulation. If an approx. = 100% modulation similar to those seen in the ultracompact candidates V407 Vu1 and Rx J0806.3+1527 were present then we would have detected it. The upper limit (3(sigma)) to any modulation at the putative orbital period is approx. 40% (rms). We extract the first X-ray spectrum from ES Cet, and find that it is not well described by simple continuum models. We find suggestive evidence for discrete spectral components at approx. 470 and 890 eV, that can be modelled as gaussian emission lines. In comparison with recent X-ray detections of nitrogen and neon in another AM CVn system (GP Com), it appears possible that these features may represent emission lines from these same elements; however, deeper spectroscopy will be required to confirm this. Our best spectral model includes a black body continuum with kT = 0.8 keV along with the gaussian lines. The 0.2 - 5 keV X-ray flux was approx. 7 x 10(exp -14) ergs/sq cm s. The luminosity implied by this flux for any reasonable distance is much smaller than that expected for a mass accretion rate as high as m = 10(exp -8) solar mass/yr, suggesting that the bulk of the accretion luminosity is below 100 eV and not seen with Chandra. We discuss the implications of our results for the nature of ES Cet.

Strohmayer, Tod E.↗

Classification of local chemical environments from x-ray absorption spectra using supervised machine learning

We report that x-ray absorption spectroscopy is a premier element-specific technique for materials characterization. Specifically, the x-ray absorption near-edge structure (XANES) encodes important information about the local chemical environment of an absorbing atom, including coordination number, symmetry, and oxidation state. Interpreting XANES spectra is a key step towards understanding the structural and electronic properties of materials, and as such, extracting structural and electronic descriptors from XANES spectra is akin to solving a challenging inverse problem. Existing methods rely on empirical fingerprints, which are often qualitative or semiquantitative and not transferable. In this paper, we present a machine learning-based approach, which is capable of classifying the local coordination environments of the absorbing atom from simulated K-edge XANES spectra. The machine learning classifiers can learn important spectral features in a broad energy range without human bias and once trained, can make predictions on the fly. The robustness and fidelity of the machine learning method are demonstrated by an average 86% accuracy across the wide chemical space of oxides in eight 3d transition-metal families. Lastly, we found that spectral features beyond the preedge region play an important role in the local structure classification problem especially for the late 3d transition-metal elements.

36 MATERIALS SCIENCE↗

Hardware Implementation of Lossless Adaptive Compression of Data From a Hyperspectral Imager

Efficient onboard data compression can reduce the data volume from hyperspectral imagers on NASA and DoD spacecraft in order to return as much imagery as possible through constrained downlink channels. Lossless compression is important for signature extraction, object recognition, and feature classification capabilities. To provide onboard data compression, a hardware implementation of a lossless hyperspectral compression algorithm was developed using a field programmable gate array (FPGA). The underlying algorithm is the Fast Lossless (FL) compression algorithm reported in Fast Lossless Compression of Multispectral- Image Data (NPO-42517), NASA Tech Briefs, Vol. 30, No. 8 (August 2006), p. 26 with the modification reported in Lossless, Multi-Spectral Data Comressor for Improved Compression for Pushbroom-Type Instruments (NPO-45473), NASA Tech Briefs, Vol. 32, No. 7 (July 2008) p. 63, which provides improved compression performance for data from pushbroom-type imagers. An FPGA implementation of the unmodified FL algorithm was previously developed and reported in Fast and Adaptive Lossless Onboard Hyperspectral Data Compression System (NPO-46867), NASA Tech Briefs, Vol. 36, No. 5 (May 2012) p. 42. The essence of the FL algorithm is adaptive linear predictive compression using the sign algorithm for filter adaption. The FL compressor achieves a combination of low complexity and compression effectiveness that exceeds that of stateof- the-art techniques currently in use. The modification changes the predictor structure to tolerate differences in sensitivity of different detector elements, as occurs in pushbroom-type imagers, which are suitable for spacecraft use. The FPGA implementation offers a low-cost, flexible solution compared to traditional ASIC (application specific integrated circuit) and can be integrated as an intellectual property (IP) for part of, e.g., a design that manages the instrument interface. The FPGA implementation was benchmarked on the Xilinx Virtex IV LX25 device, and ported to a Xilinx prototype board. The current implementation has a critical path of 29.5 ns, which dictated a clock speed of 33 MHz. The critical path delay is end-to-end measurement between the uncompressed input data and the output compression data stream. The implementation compresses one sample every clock cycle, which results in a speed of 33 Msample/s. The implementation has a rather low device use of the Xilinx Virtex IV LX25, making the total power consumption of the implementation about 1.27 W.

Keymeulen, Didlier↗

Combinations of NIR, Raman spectroscopy and physicochemical measurements for improved monitoring of solvent extraction processes using hierarchical multivariate analysis models

The reliability of chemical processes can be greatly improved by implementing inline monitoring systems. Combining multivariate analysis with non-destructive sensors can enhance the process without interfering with the operation. Here, we present here hierarchical models using both principal component analysis and partial least square analysis developed for different chemical components representative of solvent extraction process streams. A training set of 380 samples and an external validation set of 95 samples were prepared and Near infrared and Raman spectral data as well as conductivity under variable temperature conditions were collected. The results from the models indicate that careful selection of the spectral range is important. By compressing the data through Principal Component Analysis (PCA), we lower the rank of the data set to its most dominant features while maintaining the key principal components to be used in the regression analysis. Within the studied data set, concentration of five chemical components were modeled; total nitrate (NO 3 - ), total acid (H + ), neodymium (Nd 3+ ), sodium (Na + ), and ionic strength (I.S.). The best overall model prediction for each of the species studied used a combined data set comprised of complementary techniques including NIR, Raman, and conductivity. Finally, our study shows that chemometric models are powerful but requires significant amount of carefully analyzed data to capture variations in the chemistry.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Excess noise in high-current diamond diodes

We report the results of an investigation of low-frequency excess noise in high-current diamond diodes. It was found that the electronic excess noise of the diamond diodes is dominated by the 1/f and generation-recombination noise, which reveals itself as Lorentzian spectral features (f is the frequency). The generation-recombination bulges are characteristic of diamond diodes with lower turn-on voltages. The noise spectral density dependence on forward current, I, reveals three distinctive regions in all examined devices—it scales as I 2 at the low (I < 10 μA) and high (I > 10 mA) currents and, rather unusually, remains nearly constant at the intermediate current range. The characteristic trap time constants, extracted from the noise data, show a uniquely strong dependence on current. Interestingly, the performance of the diamond diodes improves with the increasing temperature. The obtained results are important for the development of noise spectroscopy-based approaches for device reliability assessment for high-power diamond electronics.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Scale-Dependent Bedform Migration and Deformation in the Physical and Spectral Domains

Multi-scale bathymetries observed in laboratory channels and natural rivers have posed several challenges to the description of bedform geometry and kinematics. To better quantify scale-dependent migration velocity, high resolution spatio-temporal bed evolution data are analyzed using three independent methods: (i) a bedform tracking method that identifies individual bedforms in longitudinal bed elevation profiles, (ii) a cross-correlation based, image matching technique in consecutive bathymetry scans similar to Particle Image Velocimetry, and (iii) two-dimensional frequency-wavenumber spectra marking a dispersion relation between bedform period and length, allowing estimation of Fourier-based, scale-dependent migration velocities. The comparative studies show that the spectral and bedform tracking methods present small differences in the migration velocity of medium-large bedforms. However, significant deviations occur in the small bedforms, mostly due to bedform sheltering or amplified exposure to the flow, which also contribute to their enhanced scale-dependent deformation. The analysis also shows that some form of scale separation is needed to extract the large and slowly migrating bedforms as they are masked by secondary features when several bedform orders coexist. Bathymetric Image Velocimetry reasonably estimates the averaged migration velocity and exhibits potential to study relations between spanwise bed surface velocity and local bedform kinematics.

58 GEOSCIENCES↗