Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “image generation”

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 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Automated defect identification in electroluminescence images of solar modules

Solar photovoltaic (PV) modules are susceptible to manufacturing defects, mishandling problems or extreme weather events that can limit energy production or cause early device failure. Trained professionals use electroluminescence (EL) images to identify defects in modules, however, field surveys or inline image acquisition can generate millions of EL images, which are infeasible to analyze by rote inspection. Here, we develop a rapid automatic computer vision pipeline (~0.5 seconds/module) to analyze EL images and identify defects including cracks, intra-cell defects, oxygen-induced defects, and solder disconnections. Defect identification is achieved with a machine learning model (Random Forest, ResNet models and YOLO) trained on 762 manually-labeled EL images of PV modules. We compare model performance on an imbalanced real-world validation set containing 134 EL images and determine that ResNet18 and YOLO are the optimal models; we next evaluated these models on a dedicated testing set (129 module images) with resulting macro F1 scores of 0.83 (ResNet18) and 0.78 (YOLO). Using a field EL survey of a PV power plant damaged in a vegetation fire, we analyze 18,954 EL images (2.4 million cells) and inspect the spatial distribution of defects on the solar modules. The results find increased frequency of ‘crack’, ‘solder’ and ‘intra-cell’ defects on the edges of the solar module closest to the ground after fire. We also find an abnormal increase of striation rings on cells which were assumed to be caused mainly in fabrication process. Our methods are published as open-source software. It can also be used to identify other kinds of defects or process different types of solar cells with minor modification on models by transfer learning.

14 SOLAR ENERGY↗
DOE OSTI2025

A case study in contrastive learning information combination: Application to technical forensics of additive manufacturing filament source identification

Combination of information from disparate data sources into a single decision is a core challenge in many fields, including the field of technical forensics. Technical forensics (TF) utilizes technical characterization of questioned samples to determine properties of that sample; these properties are then used to infer information of forensic interest, such as provenance, age, or attribution. TF is utilized in traditional forensic applications, such as the attribution of material fragments from an explosive, and in nuclear forensic applications, such as the attribution of actinides which have been interdicted out of regulatory control. The challenge of combining information from disparate sources, described alternately by many terms including “Data Fusion” and “Data Integration”, is exacerbated in the technical forensics domain due to at least two factors: the challenge of interpreting each information source singularly, and the relatively small data set sizes available. Extensive literature exists attempting to combine technical forensics information sources, both in manual and automated processes. These attempts are often bespoke to the specific information sources (such as the bi-, tri-, or quad-isotope chart (Moody, Grant, and Hutcheon 2005)), with some emerging examples of simple early- and late- fusion (, respectively). Simultaneous to the information combination efforts described in the previous paragraph, the field of natural language processing attempted (and largely succeeded) in combining information from multiple non-technical information sources. The ecosystem of “multi-modal” language models, which can take text and images as input, and generate text and images as output, became large and diverse by 2025 (Khan et al. 2025). In a generalized sense, many of these methods are trained by learning neural networks which can convert raw text or images into a vector of numbers describing the text or image, hereafter called “embeddings” and the neural networks performing the conversion are called “embedders”. By using a separate embedder for text and images, finding coincident text and images (such as images with their captions), and optimizing the parameters of the embedders such that the embeddings for the text and the image are similar, the field has found a bridge between text and images (Girdhar et al. 2023). It is the contention of the authors of this report that this insight is not limited to text and images but instead can be extended to any modality which can be found coincidently. The subject of the rest of this report is the application of this method to example multi-modal technical forensic data. Some details about the data used in this report are not appropriate for this report, and are included in a companion report (PNNL-38669).

36 MATERIALS SCIENCE↗
DOE OSTI2023

When Spectral Modeling Meets Convolutional Networks: A Method for Discovering Reionization-era Lensed Quasars in Multiband Imaging Data

Over the last two decades, around 300 quasars have been discovered at z ≳ 6, yet only one has been identified as being strongly gravitationally lensed. We explore a new approach—enlarging the permitted spectral parameter space, while introducing a new spatial geometry veto criterion—which is implemented via image-based deep learning. We first apply this approach to a systematic search for reionization-era lensed quasars, using data from the Dark Energy Survey, the Visible and Infrared Survey Telescope for Astronomy Hemisphere Survey, and the Wide-field Infrared Survey Explorer. Our search method consists of two main parts: (i) the preselection of the candidates, based on their spectral energy distributions (SEDs), using catalog-level photometry; and (ii) relative probability calculations of the candidates being a lens or some contaminant, utilizing a convolutional neural network (CNN) classification. The training data sets are constructed by painting deflected point-source lights over actual galaxy images, to generate realistic galaxy–quasar lens models, optimized to find systems with small image separations, i.e., Einstein radii of θ E ≤ 1''. Visual inspection is then performed for sources with CNN scores of P lens > 0.1, which leads us to obtain 36 newly selected lens candidates, which are awaiting spectroscopic confirmation. These findings show that automated SED modeling and deep learning pipelines, supported by modest human input, are a promising route for detecting strong lenses from large catalogs, which can overcome the veto limitations of primarily dropout-based SED selection approaches.

High-redshift galaxies↗
DOE OSTI2022

Realistic galaxy image simulation via score-based generative models

ABSTRACT We show that a denoising diffusion probabilistic model (DDPM), a class of score-based generative model, can be used to produce realistic mock images that mimic observations of galaxies. Our method is tested with Dark Energy Spectroscopic Instrument (DESI) grz imaging of galaxies from the Photometry and Rotation curve OBservations from Extragalactic Surveys (PROBES) sample and galaxies selected from the Sloan Digital Sky Survey. Subjectively, the generated galaxies are highly realistic when compared with samples from the real data set. We quantify the similarity by borrowing from the deep generative learning literature, using the ‘Fréchet inception distance’ to test for subjective and morphological similarity. We also introduce the ‘synthetic galaxy distance’ metric to compare the emergent physical properties (such as total magnitude, colour, and half-light radius) of a ground truth parent and synthesized child data set. We argue that the DDPM approach produces sharper and more realistic images than other generative methods such as adversarial networks (with the downside of more costly inference), and could be used to produce large samples of synthetic observations tailored to a specific imaging survey. We demonstrate two potential uses of the DDPM: (1) accurate inpainting of occluded data, such as satellite trails, and (2) domain transfer, where new input images can be processed to mimic the properties of the DDPM training set. Here we ‘DESI-fy’ cartoon images as a proof of concept for domain transfer. Finally, we suggest potential applications for score-based approaches that could motivate further research on this topic within the astronomical community.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2024

Toward Improving the Generation Quality of Autoregressive Slot VAEs

Unconditional scene inference and generation are challenging to learn jointly with a single compositional model. Despite encouraging progress on models that extract object-centric representations (“slots”) from images, unconditional generation of scenes from slots has received less attention. This is primarily because learning the multiobject relations necessary to imagine coherent scenes is difficult. We hypothesize that most existing slot-based models have a limited ability to learn object correlations. We propose two improvements that strengthen object correlation learning. The first is to condition the slots on a global, scene-level variable that captures higher-order correlations between slots. Second, we address the fundamental lack of a canonical order for objects in images by proposing to learn a consistent order to use for the autoregressive generation of scene objects. Specifically, we train an autoregressive slot prior to sequentially generate scene objects following a learned order. Ordered slot inference entails first estimating a randomly ordered set of slots using existing approaches for extracting slots from images, then aligning those slots to ordered slots generated autoregressively with the slot prior. Our experiments across three multiobject environments demonstrate clear gains in unconditional scene generation quality. Detailed ablation studies are also provided that validate the two proposed improvements.

MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Morphology and particle size (MaPS) exercise: testing the applications of image analysis and morphology descriptions for nuclear forensics

Image analysis techniques have been applied and shown to be a valuable tool in nuclear forensics analysis. The interlaboratory exercise reported here has tested quantitative and qualitative approaches for characterizing nuclear materials. Particle size, surface features and morphology descriptions were compared by four laboratories on a common image set generated by Scanning Electron Microscopy and Digital Light Microscopy. Quantitative analysis of the image sets through the Morphological Analysis for MAterials software highlighted the strength of image analysis, but also that the application of the software alone can introduce significant bias in the analysis. Qualitative morphology descriptions following the process outlined by Tamasi et al. (J Radioanal Nuclear Chem 307:1611–1619, 2015) were compared with a discussion on the robustness and reproducibility of the results. Finally, future work should continue to focus on proficiency and standardization of image analysis through continued exercises within the extended nuclear forensics community.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2022

SDSS IV MaNGA: visual morphological and statistical characterization of the DR15 sample

ABSTRACT We present a detailed visual morphological classification for the 4614 MaNGA galaxies in SDSS Data Release 15, using image mosaics generated from a combination of r band (SDSS and deeper DESI Legacy Surveys) images and their digital post-processing. We distinguish 13 Hubble types and identify the presence of bars and bright tidal debris. After correcting the MaNGA sample for volume completeness, we calculate the morphological fractions, the bi-variate distribution of type and stellar mass M* – where we recognize a morphological transition ‘valley’ around S0a-Sa types – and the variations of the g − i colour and luminosity-weighted age over this distribution. We identified bars in 46.8 per cent of galaxies, present in all Hubble types later than S0. This fraction amounts to a factor ∼2 larger when compared with other works for samples in common. We detected 14 per cent of galaxies with tidal features, with the fraction changing with M* and morphology. For 355 galaxies, the classification was uncertain; they are visually faint, mostly of low/intermediate masses, low concentrations, and discy in nature. Our morphological classification agrees well with other works for samples in common, though some particular differences emerge, showing that our image procedures allow us to identify a wealth of added value information as compared to SDSS-based previous estimates. Based on our classification, we also propose an alternative criteria for the E–S0 separation, in the structural semimajor to semiminor axis versus bulge to total light ratio (b/a − B/T) and concentration versus semimajor to semiminor axis (C − b/a) space.

Vázquez-Mata, J. A. (ORCID:0000000186941204)↗
DOE OSTI2020

A new generation of direct X-ray detectors for medical and synchrotron imaging applications

Abstract Large-area X-ray imaging is one of the most widely used imaging modalities that spans several scientific and technological fields. Currently, the direct X-ray conversion materials that are being commercially used for large-area (> 8 cm × 4 cm without tiling) flat panel applications, such as amorphous selenium (a-Se), have usable sensitivities of up to only 30 keV. Although there have been many promising candidates (such as polycrystalline HgI 2 and CdTe), none of the semiconductors were able to assuage the requirement for high energy (> 40 keV) large-area X-ray imaging applications due to inadequate cost, manufacturability, and long-term performance metrics. In this study, we successfully demonstrate the potential of the hybrid Methylammonium lead iodide (MAPbI 3 ) perovskite-based semiconductor detectors in satisfying all the requirements for its successful commercialization in synchrotron and medical imaging. This new generation of hybrid detectors demonstrates low dark current under electric fields needed for high sensitivity X-ray imaging applications. The detectors have a linear response to X-ray energy and applied bias, no polarization effects at a moderate bias, and signal stability over long usage durations. Also, these detectors have demonstrated a stable detection response under BNL’s National Synchrotron Light Source II (NSLS-II) 70 keV monochromatic synchrotron beamline.

47 OTHER INSTRUMENTATION↗
DOE OSTI2022

X-ray morphology of cluster-mass haloes in self-interacting dark matter

ABSTRACT We perform cosmological zoom-in simulations of 19 relaxed cluster-mass haloes with the inclusion of adiabatic gas in the cold dark matter (CDM) and self-interacting dark matter (SIDM) models. These clusters are selected as dynamically relaxed clusters from a parent simulation with $M_{\rm 200} \simeq (1\!-\!3)\times 10^{15}{\, \rm M_\odot }$. Both the dark matter and the intracluster gas distributions in SIDM appear more spherical than their CDM counterparts. Mock X-ray images are generated based on the simulations and are compared to the real X-ray images of 84 relaxed clusters selected from the Chandra and ROSAT archives. We perform ellipse fitting for the isophotes of mock and real X-ray images and obtain the ellipticities at cluster-centric radii of $r\simeq 0.1\!-\!0.2R_{\rm 200}$. The X-ray isophotes in SIDM models with increasing cross-sections are rounder than their CDM counterparts, which manifests as a systematic shift in the distribution function of ellipticities. Unexpectedly, the X-ray morphology of the observed non-cool-core clusters agrees better with SIDM models with cross-section $(\sigma /m)= 0.5\!-\!1\, {\rm cm}^2\, {\rm g}^{-1}$ than CDM and SIDM with $(\sigma /m)=0.1\, {\rm cm}^2\, {\rm g}^{-1}$. Our statistical analysis indicates that the latter two models are disfavoured at the $68{{\ \rm per\ cent}}$ confidence level (as conservative estimates). This conclusion is not altered by shifting the radial range of measurements or applying a temperature selection criterion. However, the primary uncertainty originates from the lack of baryonic physics in the adiabatic model, such as cooling, star formation and feedback effects, which still have the potential to reconcile CDM simulations with observations.

Shen, Xuejian (ORCID:000000026196823X)↗
DOE OSTI2025

Production and radiochemistry of the in vivo PET generator 140 Nd/ 140 Pr as an imaging surrogate for F-block therapeutic radionuclides

Theranostics, a combined approach of diagnostics and therapeutics, often employs F-block therapeutic radionuclides including 225 Ac, 177 Lu, and 161 Tb. While there is a lack of F-block PET imaging radionuclides, the in vivo PET generator pair 140 Nd/ 140 Pr can act as a theranostic imaging counterpart to the F-block therapeutic radionuclides. In this study, we explored the production and separation of high purity 140 Nd via the 141 Pr(p,2n) 140 Nd reaction route. Monoisotopic 141 Pr targets irradiated with 20 MeV protons for 10 min with 10 µA beam current yielded 21.45 ± 0.82 MBq (580 ± 22 µCi) of 140 Nd. A two-step separation method was developed for the purification of 140 Nd from the 141 Pr target material. Recoveries of 27.4 ± 2.1% 140Nd were obtained upon separation with < 20 ppb of 141 Pr target material in the final product. Radiolabeling of Macropa and DOTA chelators with 140 Nd resulted in [ 140 Nd]Nd-Macropa with a molar activity of 74.0 MBq/µmol (2.0 mCi/µmol) and [ 140 Nd]Nd-DOTA with a molar activity of 70.3 MBq/µmol (1.9 mCi/µmol). An imaging study with a phantom indicated the PET spatial resolution of 140 Nd/ 140 Pr was distinguishable down to 2.4 mm. This study sets the stage for the 140 Nd/ 140 Pr in vivo PET generator to be explored in radiopharmaceutical applications.

F-block↗
DOE OSTI2026

Dual particle imaging using time-of-flight neutron classification

Fast-neutron imaging technology is well-suited for passive nuclear material monitoring, secondary inspection of flagged cargo, and wide-area search for lost neutron sources. However, imaging systems that use pulse shape discrimination for event classification require complex pulse waveform analysis. In this work, we evaluate time-of-flight (TOF) based particle classification as an alternative solution for fast-neutron imaging by classifying all events with a TOF above a maximum threshold as neutrons. We measured a Cf-252 source next to Cs-137 using a 12-bar organic-glass scintillator array. By varying the TOF thresholds for neutron identification, we demonstrate a clear trade-off between event yield and backprojection image fidelity, with stricter thresholds improving precision at the cost of statistics, TOF thresholded data generated an image that predicted the neutron source direction with 20% reduced mean central angle prediction error compared to a traditional pulse shape discrimination (PSD) method with comparable event count. Time-of-flight particle classification shows promise as an alternative to pulse shape discrimination systems for fast neutron imaging systems looking to minimize costs and size of electronics with comparable imaging quality. The sources used demonstrate that the method is effective in classifying measured neutrons in a measurement environment with 150 μCi Cs-137 and 1.6 × 10 6 n/s Cf-252 sources positioned at distances of 66 cm and 81 cm from the detector. Additionally, the method classifies low-energy neutron events that pulse shape discrimination removes, so a combination of both methods would result in a higher overall neutron event efficiency.

Heriot, William [Univ. of Michigan, Ann Arbor, MI ↗
DOE OSTI2021

Computer-aided Abnormality Detection in Chest Radiographs in a Clinical Setting via Domain-adaptation

Deep learning (DL) models are being deployed at medical centers to aid radiologists for diagnosis of lung conditions from chest radiographs. Such models are often trained on a large volume of publicly available labeled radiographs. These pre-trained DL models’ ability to generalize in clinical settings is poor because of the changes in data distributions between publicly available and privately held radiographs. In chest radiographs, the heterogeneity in distributions arises from the diverse conditions in X-ray equipment and their configurations used for generating the images. In the machine learning community, the challenges posed by the heterogeneity in the data generation source is known as domain shift, which is a mode shift in the generative model. In this work, we introduce a domain-shift detection and removal method to overcome this problem. Our experimental results show the proposed method’s effectiveness in deploying a pre-trained DL model for abnormality detection in chest radiographs in a clinical setting.

Dubey, Abhishek↗
DOE OSTI2020

Neutron sub-micrometre tomography from scattering data

Neutrons are valuable probes for various material samples across many areas of research. Neutron imaging typically has a spatial resolution of larger than 20 µm, whereas neutron scattering is sensitive to smaller features but does not provide a real-space image of the sample. A computed-tomography technique is demonstrated that uses neutron-scattering data to generate an image of a periodic sample with a spatial resolution of ∼300 nm. The achieved resolution is over an order of magnitude smaller than the resolution of other forms of neutron tomography. This method consists of measuring neutron diffraction using a double-crystal diffractometer as a function of sample rotation and then using a phase-retrieval algorithm followed by tomographic reconstruction to generate a map of the sample's scattering-length density. Topological features found in the reconstructions are confirmed with scanning electron micrographs. This technique should be applicable to any sample that generates clear neutron-diffraction patterns, including nanofabricated samples, biological membranes and magnetic materials, such as skyrmion lattices.

nanoscience↗
DOE OSTI2025

Continuous-variable quantum Boltzmann machine

Here, we propose a continuous-variable quantum Boltzmann machine (CVQBM) using a powerful energy-based neural network. It can be realized experimentally on a continuous-variable (CV) photonic quantum computer. We used a CV quantum imaginary time evolution (QITE) algorithm to prepare the essential thermal state and then designed the CVQBM to proficiently generate continuous probability distributions. We applied our method to both classical and quantum data. Using real-world classical data, such as synthetic-aperture radar (SAR) images, we generated probability distributions. For quantum data, we used the output of CV quantum circuits. We obtained high fidelity and low Kullback–Leibler (KL) divergence showing that our CVQBM learns distributions from given data well and generates data sampling from that distribution efficiently. We also discussed the experimental feasibility of our proposed CVQBM. Our method can be applied to a wide range of real-world problems by choosing an appropriate target distribution (corresponding to, e.g., SAR images, medical images, and risk management in finance). Moreover, our CVQBM is versatile and could be programmed to perform tasks beyond generation, such as anomaly detection.

SAR images↗
DOE OSTI2020

Sampling Subjective Polygons for Patch-Based Deep Learning Land-Use Classification in Satellite Images

Model generalization remains a key challenge in the analysis of large amounts of heterogeneous satellite image data. One major limiting factor in developing generalizable models, in the context of supervised learning, is the lack of high quality training datasets. A model's capacity to perform well on new data is often inhibited by imbalance and bias in the data that was used for training. This is especially a problem when using convolutional neural networks to classify urban land-use in satellite images. Notable dataset imbalance issues in this application include land-use type imbalance and image scene imbalance. To begin understanding these dataset imbalance problems in more detail, we develop and test a number of sampling methods for generating training image datasets from subjective training polygons for urban land-use classification. We investigate sampling at different point densities as a means to reduce content repetition and therefore content imbalance and bias in the training image dataset.

Arndt, Jacob↗
DOE OSTI2021

Scalable FBP decomposition for cone-beam CT reconstruction

Filtered Back-Projection (FBP) is a fundamental compute intense algorithm used in tomographic image reconstruction. Cone-Beam Computed Tomography (CBCT) devices use a cone-shaped X-ray beam, in comparison to the parallel beam used in older CT generations. Distributed image reconstruction of cone-beam datasets typically relies on dividing batches of images into different nodes. This simple input decomposition, however, introduces limits on input/output sizes and scalability.We propose a novel decomposition scheme and reconstruction algorithm for distributed FPB. This scheme enables arbitrarily large input/output sizes, eliminates the redundancy arising in the end-to-end pipeline and improves the scalability by replacing two communication collectives with only one segmented reduction. Finally, we implement the proposed decomposition scheme in a framework that is useful for all current-generation CT devices (7th gen). In our experiments using up to 1024 GPUs, our framework can construct 40963 volumes, for real-world datasets, in under 16 seconds (including I/O).

Chen, Peng↗
DOE OSTI2020

Non-invasive (non-contact) measurements of human thermal physiology signals and thermal comfort/discomfort poses -A review

Heating, ventilation and air-conditioning (HVAC) systems have been adopted to create comfortable, healthy and safe indoor environments. In the control loop, the technical feature of the human demand-oriented supply can help operate HVAC effectively. Among many technical options, real time monitoring based on feedback signals from end users has been frequently reported as a critical technology to confirm optimizing building performance. Recent studies have incorporated human thermal physiology signals and thermal comfort/discomfort status as real-time feedback signals. A series of human subject experiments used to be conducted by primarily adopting subjective questionnaire surveys in a lab-setting study, which is limited in the application for reality. With the help of advanced technologies, physiological signals have been detected, measured and processed by using multiple technical formats, such as wearable sensors. Nevertheless, they mostly require physical contacts with the skin surface in spite of the small physical dimension and compatibility with other wearable accessories, such as goggles, and intelligent bracelets. Most recently, a low cost small infrared camera has been adopted for monitoring human facial images, which could detect the facial skin temperature and blood perfusion in a contactless way. Also, according to latest pilot studies, a conventional digital camera can generate infrared images with the help of new methods, such as the Euler video magnification technology. Human thermal comfort/discomfort poses can also be detected by video methods without contacting human bodies and be analyzed by the skeleton keypoints model. Here, new sensing technologies were summarized, their cons and pros were discussed, and extended applications for the demand-oriented ventilation were also reviewed as potential development and applications.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗