Image Similarity Search Poster
Empowering Nuclear Energy Research with Advanced Image Similarity Search
97 - MATHEMATICS AND COMPUTING↗
SEARCH · Search NASA
Results for “Image Similarity”
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.
AI/ML Expo Boosting Job Performance with AI: Innovative Approaches and Success Stories
Our technology leverages artificial intelligence (AI) to enhance the user experience in High Performance Computing (HPC) environments. By analyzing user behavior and providing personalized recommendations, our AI system helps HPC users optimize their workflows and improve productivity. Additionally, we offer an advanced image similarity search feature, which utilizes AI algorithms to identify and retrieve visually similar images, saving users valuable time and effort in their research and analysis.
97 - MATHEMATICS AND COMPUTING↗
Comparison of Polarized Radiative Transfer Codes Used by the EHT Collaboration
Interpretation of resolved polarized images of black holes by the Event Horizon Telescope (EHT) requires predictions of the polarized emission observable by an Earth-based instrument for a particular model of the black hole accretion system. Such predictions are generated by general relativistic radiative transfer (GRRT) codes, which integrate the equations of polarized radiative transfer in curved spacetime. A selection of ray-tracing GRRT codes used within the EHT Collaboration is evaluated for accuracy and consistency in producing a selection of test images, demonstrating that the various methods and implementations of radiative transfer calculations are highly consistent. When imaging an analytic accretion model, we find that all codes produce images similar within a pixel-wise normalized mean squared error (NMSE) of 0.012 in the worst case. When imaging a snapshot from a cell-based magnetohydrodynamic simulation, we find all test images to be similar within NMSEs of 0.02, 0.04, 0.04, and 0.12 in Stokes I , Q , U , and V , respectively. We additionally find the values of several image metrics relevant to published EHT results to be in agreement to much better precision than measurement uncertainties.
79 ASTRONOMY AND ASTROPHYSICS↗
Galaxy deblending using residual dense neural networks
We present a new neural network approach for deblending galaxy images in astronomical data using Residual Dense Neural network (RDN) architecture. We train the network on synthetic galaxy images similar to the typical arrangements of field galaxies with a finite point spread function (PSF) and realistic noise levels. The main novelty of our approach is the usage of two distinct neural networks: i) a deblending network which isolates a single galaxy postage stamp from the composite and, ii) a classifier network which counts the remaining number of galaxies. The deblending proceeds by iteratively peeling one galaxy at a time from the composite until the image contains no further objects as determined by the classifier, or by other stopping criteria. By looking at the consistency in the outputs of the two networks, we can assess the quality of the deblending. Further, we characterize the flux and shape reconstructions in different quality bins and compare our deblender with the industry standard, SExtractor. We also discuss possible future extensions for the project with variable PSFs and noise levels.
79 ASTRONOMY AND ASTROPHYSICS↗
Investigation of near-field jet stability of a single-hole injector based on fast X-ray phase contrast imaging and image feature matching
Fuel spray is very effective in controlling the combustion process to improve engine performance and reduce emissions. Understanding the spray unstability during the injection process is of importance to improve the control of spray characteristics and engine operation. In this study, near-field biodiesel jets were recorded using fast X-ray phase contrast imaging and the flow features inside the jet were extracted using Speeded Up Robust Features (SURF) method. Here, the image similarity by feature matching was successfully used to represent the near-field jet stability. Based on the jet stability, an injection process can be divided into five stages: an unstable stage at needle opening, a partially stable transition stage at needle opening, a stable stage, a stable transition stage at needle closing and an unstable stage at needle closing. The ranges of needle lift for five stages were also determined. The jet unstability is highly related to the cavitation formation and gas purging process during needle opening. The variation of stable jet feature is dependent on the needle lift at needle closing. Higher needle lift for the similar jet feature at needle opening indicates a hydraulic delay compared to needle closing. Finally, the possible reasons of jet feature formation and feature detection used on the multi-hole injector are discussed.
33 ADVANCED PROPULSION SYSTEMS↗
Special Nuclear Material Hold-up Measurement
Facilities that process special nuclear material (SNM) generally have a variety of equipment, shielding, and nearby radiation sources, which can pose challenges when performing nondestructive analysis (NDA) within the facility, as shown in Figure 1. In this figure, the left image features two individuals performing a holdup measurement of a large duct at a posting, utilizing a detector probe attached to a yardstick to reduce some aspect of the measurement uncertainty. Note that the distance to the holdup within the pipe from the posting cannot be accounted for. The right image similarly features several individuals guiding a holdup measurement of a ducting pipe several meters overhead, by attaching a detector probe to a long handle and lining that up with a posting on the outside of the pipe. SNM processing begets SNM holdup, and this difficulty with accurate NDA can result in unaccounted accumulation of SNM holdup within a process or area. Normal SNM processing operations can accumulate holdup within filters, pumps, pipes, ducts, other equipment, and facility support systems, as demonstrated by Figure 2. It is important for deactivation and decommissioning (D&D) gloveboxes and their ancillary equipment, material accountability, criticality safety, facility operations, waste management, radiation safety, and security to accurately characterize the location, composition, and quantity of this holdup to minimize these associated risks and potential diversion pathways. Current NDA techniques for assay of holdup are expensive, time consuming, and have high uncertainty.
11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
Radioimaging for real-time tracking of high-voltage breakdown
Development of a radioimaging diagnostic for high-voltage component reliability testing and electrical breakdown computational model validation is described. Radioimaging has its roots in radio astronomy, where aperture synthesis (also known as synthesis imaging) has been utilized for decades to image radio sources far from Earth. Radioimaging as described herein, in contrast, seeks to image radio sources in close proximity to its receivers (i.e., in a laboratory environment). Here it is shown that corona discharge, a non-destructive precursor to catastrophic (thermal) arc discharge, electromagnetically radiates strongly within a 250 kHz – 2.5 GHz bandwidth, and is readily detected and located by postprocessing the received radio signals. The ability of radioimaging to detect both corona and arc discharge (grouped together herein as high voltage breakdown or HVB) makes it a valuable tool for 100% HVB detection in materials, components, and devices, and has the ability to indicate electrical weakness (via corona detection) prior to a destructive arc discharge event. Radioimaging enables HVB to be located both internal and external to dielectric components under test in near-real-time, with multiple and/or extended HVB events located simultaneously. In contrast, existing non-destructive diagnostics (at the time of this writing) either indicate electrical breakdown without resolving failure locations (e.g., current, voltage, and chemical measurements), locate external HVB (e.g., high-speed optical and ultraviolet (UV) measurements or photography), or locate both external and internal HVB but with low fidelity (e.g., a single HVB source can be located by existing time-of-arrival (TOA) UHF or acoustic emissions). Radioimaging instead creates a sequence of high-fidelity images similar to an optical high-speed camera but at radiofrequencies (RF), and is not limited to two-dimensions. Moreover, radioimaging has already served one internal and two external industry customers, the results of which are detailed in this report. The radioimaging results described herein were part of a three-year effort funded by the Sandia Lab Directed Research and Development (LDRD) program within the Radiation, Electromagnetic, High Energy Density Science (REHEDS) investment area.
47 OTHER INSTRUMENTATION↗
Batch Active Learning for Multispectral and Hyperspectral Image Segmentation Using Similarity Graphs
Abstract Graph learning, when used as a semi-supervised learning (SSL) method, performs well for classification tasks with a low label rate. We provide a graph-based batch active learning pipeline for pixel/patch neighborhood multi- or hyperspectral image segmentation. Our batch active learning approach selects a collection of unlabeled pixels that satisfy a graph local maximum constraint for the active learning acquisition function that determines the relative importance of each pixel to the classification. This work builds on recent advances in the design of novel active learning acquisition functions (e.g., the Model Change approach in arXiv:2110.07739) while adding important further developments including patch-neighborhood image analysis and batch active learning methods to further increase the accuracy and greatly increase the computational efficiency of these methods. In addition to improvements in the accuracy, our approach can greatly reduce the number of labeled pixels needed to achieve the same level of the accuracy based on randomly selected labeled pixels.
97 MATHEMATICS AND COMPUTING↗
Plasma image classification using cosine similarity constrained convolutional neural network
Plasma jets are widely investigated both in the laboratory and in nature. Astrophysical objects such as black holes, active galactic nuclei and young stellar objects commonly emit plasma jets in various forms. With the availability of data from plasma jet experiments resembling astrophysical plasma jets, classification of such data would potentially aid in not only investigating the underlying physics of the experiments but also the study of astrophysical jets. In this work we use deep learning to process all of the laboratory plasma images from the Caltech Spheromak Experiment spanning two decades. We found that cosine similarity can aid in feature selection, classify images through comparison of feature vector direction and be used as a loss function for the training of AlexNet for plasma image classification. We also develop a simple vector direction comparison algorithm for binary and multi-class classification. Using our algorithm we demonstrate 93 % accurate binary classification to distinguish unstable columns from stable columns and 92 % accurate five-way classification of a small, labelled data set which includes three classes corresponding to varying levels of kink instability.
70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
ESM data downscaling: a comparison of super-resolution deep learning models
Abstract Climate projections at fine spatial resolutions are required to conduct accurate risk assessment for critical infrastructure and design adaptation planning. Generating these projections using advanced Earth system models (ESM) requires significant computational resources. To address this issue, various statistical downscaling techniques have been introduced to generate fine-resolution data from coarse-resolution simulations. In this study, we evaluate and compare five deep learning-based downscaling techniques, namely, super-resolution convolutional neural networks, fast super-resolution convolutional neural network ESM, efficient sub-pixel convolutional neural network, enhanced deep residual network (EDRN), and super-resolution generative adversarial network (SRGAN). These techniques are applied to a dataset generated by the Energy Exascale Earth System Model (E3SM), focusing on key surface variables such as surface temperature, shortwave heat flux, and longwave heat flux. Models are trained and validated using paired fine-resolution (0.25 $$^{\circ }$$ ∘ ) and coarse-resolution (1 $$^{\circ }$$ ∘ ) monthly data obtained from a 9-year simulation. Next, blind testing is performed using monthly data obtained from two different years outside of the training and validation set. To evaluate the efficiency of each technique, different statistical metrics are used, including mean squared error (MSE), peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and learned perceptual image patch similarity (LPIPS). The results show that EDRN outperforms other algorithms in terms of PSNR, SSIM, and MSE, but struggles to capture fine-scale features in the data. In contrast, SRGAN, a generative model that uses perceptual loss, excels in capturing fine details at boundaries and internal structures, resulting in lower LPIPS than other methods.
Pawar, Nikhil M. (ORCID:0000000211613289)↗
Concept Lens: Visual Comparison and Evaluation of Generative Model Manipulations
Generative models are becoming a transformative technology for the creation and editing of images. However, it remains challenging to harness these models for precise image manipulation. These challenges often manifest as inconsistency in the editing process, where both the type and amount of semantic change, depend on the image being manipulated. Moreover, there exist many methods for computing image manipulations, whose development is hindered by the matter of inconsistency. This paper aims to address these challenges by improving how we evaluate, compare, and explore the space of manipulations offered by a generative model. We present Concept Lens, a visual interface that is designed to aid users in understanding semantic concepts carried in image manipulations, and how these manipulations vary over generated images. Given the large space of possible images produced by a generative model, Concept Lens is designed to support the exploration of both generated images, and their manipulations, at multiple levels of detail. To this end, the layout of Concept Lens is informed by two hierarchies: a hierarchical organization of (1) original images, grouped by their similarities, and (2) image manipulations, where manipulations that induce similar changes are grouped together. This layout allows one to discover the types of images that consistently respond to a group of manipulations, and vice versa, manipulations that consistently respond to a group of codes. We show the benefits of this design across multiple use cases, specifically, studying the quality of manipulations for a single method, and offering a means of comparing different methods.
clustering↗
Comparing Deep Learning Approaches for Understanding Genotype × Phenotype Interactions in Biomass Sorghum
We explore the use of deep convolutional neural networks (CNNs) trained on overhead imagery of biomass sorghum to ascertain the relationship between single nucleotide polymorphisms (SNPs), or groups of related SNPs, and the phenotypes they control. We consider both CNNs trained explicitly on the classification task of predicting whether an image shows a plant with a reference or alternate version of various SNPs as well as CNNs trained to create data-driven features based on learning features so that images from the same plot are more similar than images from different plots, and then using the features this network learns for genetic marker classification. We characterize how efficient both approaches are at predicting the presence or absence of a genetic markers, and visualize what parts of the images are most important for those predictions. We find that the data-driven approaches give somewhat higher prediction performance, but have visualizations that are harder to interpret; and we give suggestions of potential future machine learning research and discuss the possibilities of using this approach to uncover unknown genotype × phenotype relationships.
Zhang, Zeyu↗
First graded metal pushered single shell capsule implosions on the National Ignition Facility
Graded metal pushered single shell (PSS) capsules are predicted to be a viable alternative to low-Z capsule indirect drive inertial confinement fusion (ICF) implosions for achieving high fusion yields [MacLaren et al., Phys. Plasmas 28, 122710 (2021)]. The first experiments with Be/Cr-graded metal PSS capsules indicate that the implementation of the principle design feature, the graded density inner metal layer, has succeeded in producing a stable implosion with performance in agreement with predictions. With 50% Cr concentration in the pusher, PSS capsules have greater than ~2× higher shell densities during stagnation for enhanced core confinement and radiation trapping at ~35% lower shell implosion velocities than low-Z capsules. High-energy >30 keV inflight shell radiography recorded 215 km/s implosion velocities and show that implosion Legendre mode P 2 asymmetry can be tuned via inner-to-outer beam wavelength separation, similar to other implosions. Shell radiographs and neutron core images show similar P 2 asymmetry, suggesting no symmetry swings between peak implosion velocity and stagnation times. Despite the modest implosion velocities, gas-filled deuterium–tritium capsule implosions generate 10 15 neutron yields at relatively modest core ion temperatures of 2.75 keV, indicating that in spite of the high-density inner layer, the implosions have been stabilized by the design density gradient. When compared with hydrodynamic simulations, the measured yield-over-simulated is 35% due to fuel–pusher mix and other perturbations such as the capsule fill tube. Finally, simple analytical scalings of hot spot pressure and neutron yield show that PSS implosions reach similar performance at lower implosion velocities and higher shell densities to low-Z ICF capsules.
70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
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↗
SNIa Cosmology Analysis Results from Simulated LSST Images: From Difference Imaging to Constraints on Dark Energy
Abstract The Vera Rubin Observatory Legacy Survey of Space and Time (LSST) is expected to process ∼10 6 transient detections per night. For precision measurements of cosmological parameters and rates, it is critical to understand the detection efficiency, magnitude limits, artifact contamination levels, and biases in the selection and photometry. Here we rigorously test the LSST Difference Image Analysis (DIA) pipeline using simulated images from the Rubin Observatory LSST Dark Energy Science Collaboration Data Challenge (DC2) simulation for the Wide-Fast-Deep survey area. DC2 is the first large-scale (300 deg 2 ) image simulation of a transient survey that includes realistic cadence, variable observing conditions, and CCD image artifacts. We analyze ∼15 deg 2 of DC2 over a 5 yr time span in which artificial point sources from Type Ia supernova (SNIa) light curves have been overlaid onto the images. The magnitude limits per filter are u = 23.66 mag, g = 24.69 mag, r = 24.06 mag, i = 23.45 mag, z = 22.54 mag, and y = 21.62 mag. The artifact contamination levels are ∼90% of all detections, corresponding to ∼1000 artifacts deg –2 in g band, and falling to 300 deg –2 in y band. The photometry has biases <1% for magnitudes 19.5 < m < 23. Our DIA performance on simulated images is similar to that of the Dark Energy Survey difference-imaging pipeline on real images. We also characterize DC2 image properties to produce catalog-level simulations needed for distance bias corrections. We find good agreement between DC2 data and simulations for distributions of signal-to-noise ratio, redshift, and fitted light-curve properties. Applying a realistic SNIa cosmology analysis for redshifts z < 1, we recover the input cosmology parameters to within statistical uncertainties.
79 ASTRONOMY AND ASTROPHYSICS↗
Nuclear modification of $\Upsilon$ states in pPb collisions at $\sqrt{s_\mathrm{NN}}$ = 5.02 TeV
Production cross sections of Image 1, Image 2, and Image 3 states decaying into Image 4 in proton-lead ( p Pb ) collisions are reported using data collected by the CMS experiment at s NN = 5.02 TeV . A comparison is made with corresponding cross sections obtained with p p data measured at the same collision energy and scaled by the Pb nucleus mass number. The nuclear modification factor for Image 1 is found to be Image 5. Similar results for the excited states indicate a sequential suppression pattern, such that Image 6. The suppression of all states is much less pronounced in p Pb than in PbPb collisions, and independent of transverse momentum Image 7 and center-of-mass rapidity Image 8 of the individual Image 9 state in the studied range Image 10 and Image 11. Models that incorporate final-state effects of bottomonia in pPb collisions are in better agreement with the data than those which only assume initial-state modifications.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗