SEARCH · Search NASA
Results for “Feature detection”
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.
Improving Non-Destructive Detection Technology Through SAVY Feature Detection
Surveillance of special nuclear material (SNM) storage containers is required by the DOE to assess their integrity across the Complex. This work aims to improve the task of container inspections by leveraging automation through machine learning (ML) tools to reduce the human-intensive effort and expert-level knowledge needed to assess container status. A field-deployable, non-destructive technology was designed using off-the-shelf components to collect multiple images from different perspectives of containers in storage to detect both spatial features of interest and anomalies of concern. Nine ML models were generated using unique training datasets and parameters. Learned features include SAVY surface regions including the body side wall, collar, lid, filter, and printed/etched information. Average Precision (AP) is used to calculate detection performance when both viewing previously seen environments and previously unseen environments. The application of image transformations and resolution scaling while training greatly improved the detection performance in unseen environments, and significantly increasing the number of computation iterations improved detection performance on previously seen environments. Additional capabilities were developed including the novel detection of procedural non-compliance and the ability to localize anomalies relative to SAVY surface features.
A detected feature at the expected wavelength for the HD R5(0) line in Jupiter's and Uranus' atmospheres
A feature was detected at the expected wavelength for the HD R5(0) line in Jupiter's and Uranus' atmospheres. An upper limit for Neptune was also found. Added to the earlier detection of a similar feature for Saturn, it is proposed that all evidence from this type of measurement can be interpreted as arising from a D/H ratio of about .0004 for all the major planets. This value is not in agreement with measurements from CH3D transitions, and is at least fifty times the accepted interstellar medium value of 5 x .000001, implying deuterium enhancement in the solar system via fractionation in the proto-solar nebula.
Synthetic aperture radar target detection, feature extraction, and image formation techniques
This report presents new algorithms for target detection, feature extraction, and image formation with the synthetic aperture radar (SAR) technology. For target detection, we consider target detection with SAR and coherent subtraction. We also study how the image false alarm rates are related to the target template false alarm rates when target templates are used for target detection. For feature extraction from SAR images, we present a computationally efficient eigenstructure-based 2D-MODE algorithm for two-dimensional frequency estimation. For SAR image formation, we present a robust parametric data model for estimating high resolution range signatures of radar targets and for forming high resolution SAR images.
Feature Detection
Focal Area(s): This proposal aims to develop and evaluate statistical models and machine learning algorithms for detecting and tracking features in spatiotemporal remotely sensed data with uncertainty quantification. We focus a particular application on the detection of sea ice leads and ridges in the Arctic and use these key sea ice features for model calibration and to gain insight into the physics of sea ice thermodynamics and deformation.
TempestExtremes v2.1: a community framework for feature detection, tracking, and analysis in large datasets
TempestExtremes (TE) is a multifaceted framework for feature detection, tracking, and scientific analysis of regional or global Earth system datasets on either rectilinear or unstructured/native grids. Version 2.1 of the TE framework now provides extensive support for examining both nodal (i.e., pointwise) and areal features, including tropical and extratropical cyclones, monsoonal lows and depressions, atmospheric rivers, atmospheric blocking, precipitation clusters, and heat waves. Available operations include nodal and areal thresholding, calculations of quantities related to nodal features such as accumulated cyclone energy and azimuthal wind profiles, filtering data based on the characteristics of nodal features, and stereographic compositing. This paper describes the core algorithms (kernels) that have been added to the TE framework since version 1.0, including algorithms for editing pointwise trajectory files, composition of fields around nodal features, generation of areal masks via thresholding and nodal features, and tracking of areal features in time. Several examples are provided of how these kernels can be combined to produce composite algorithms for evaluating and understanding common atmospheric features and their underlying processes. These examples include analyzing the fraction of precipitation from tropical cyclones, compositing meteorological fields around extratropical cyclones, calculating fractional contribution to poleward vapor transport from atmospheric rivers, and building a climatology of atmospheric blocks.
LIC for Surface Flow Feature Detection
The Line Integral Convolution (LIC) algorithm has received a lot of attention and interest. Yet, only a few of the current LIC related algorithms deal specifically with color textures for automatic detection of flow features. This paper provides an overview of research in this area.
Aerosol and Cloud Optical Properties from the ARM Raman Lidars: The Feature Detection and Extinction (RLPROF-FEX) Value-Added Product
Aerosols and their interactions and influence on clouds are among the main sources of uncertainties in radiative direct and indirect forcing (IPCC 2013). Continuous height-resolved measurements of cloud and aerosol optical properties are needed to reduce these uncertainties. Here we describe the Raman Lidar Profiles – Feature detection and Extinction (RLPROF-FEX) Value-Added Product (VAP) derived using Raman lidar data at multiple U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) user facility sites. RLPROF-FEX provides estimates of extinction, backscatter, and depolarization using the algorithm described by Thorsen et al. 2015 and Thorsen and Fu 2015. This document provides a description of the FEX algorithm, its input and output data, and related details about the Raman lidar (RL) system.
The use of linear feature detection to investigate thematic mapper data performance and processing
The geometric and radiometric characteristics of thematic mapper data through analysis of linear features in the data are investigated. The particular aspects considered are: (1) thematic mapper ground IFOV; (2) radiometric contrast between linear features and background; (3) precision of system geometric correction; (4) band-to-band registration; and (5) potential utility of TM data for linear feature detection especially as compared to MSS data. It is shown that TM data may be used to estimate TM pixel size and to illustrate band-band mis-registration. Further, the geometry and radiometry of the data are sufficiently precise to allow accurate estimation of the widths of linear features. In optimum conditions features one quarter of a pixel in width may be accurately measured. These results have considerable potential for applications for hydrological and topographic mapping.
The Use of Linear Feature Detection to Investigate Thematic Mapper Data Performance and Processing
The geometric and radiometric characteristics of thematic mapper data through analysis of linear features in the data are investigated. The particular aspects considered are: (1) thematic mapper ground IFUV; (2) radiometric contrast between linear features and background; (3) precision of system geometric correction; (4) band-to-band registration; and (5) potential utility of TM data for linear feature detection especially as compared to MSS data. It is shown that TM data may be used to estimate TM pixel size illustrate band: band mis-registration.
Feature Detection in Linked Derived Spaces
This paper describes by example a strategy for plotting and interacting with data in multiple metric spaces. The example system was designed for use with time-varying computational fluid dynamics (CFD) datasets, but the methodology is directly applicable to other types of field data. The central objects embodied by the tool are {\em portraits}, which show the data in various coordinate systems, while preserving their spatial connectivity and temporal variability. The coordinates are derived in various ways from the field data, and an important feature is that new and derived portraits can be created interactively. The primary operations supported by the tool are brushing and linking: the user can select a subset of a given portrait, and this subset is highlighted in all portraits. The user can combine highlighted subsets from an arbitrary number of portraits with the usual logical operators, thereby indicating where an arbitrarily complex set of conditions holds. The system is useful for exploratory visualization and feature detection in multivariate data.
CATS Version 2 Aerosol Feature Detection and Applications for Data Assimilation
The Cloud Aerosol Transport System (CATS) lidar has been operating onboard the International Space Station (ISS) since February 2015 and provides vertical observations of clouds and aerosols using total attenuated backscatter and depolarization measurements. From February March 2015, CATS operated in Mode 1, providing backscatter and depolarization measurements at 532 and 1064 nm. CATS began operation in Mode 2 in March 2015, providing backscatter and depolarization measurements at 1064 nm and has continued to operate to the present in this mode. CATS level 2 products are derived from these measurements, including feature detection, cloud aerosol discrimination, cloud and aerosol typing, and optical properties of cloud and aerosol layers. Here, we present changes to our level 2 algorithms, which were aimed at reducing several biases in our version 1 level 2 data products. These changes will be incorporated into our upcoming version 2 level 2 data release in summer 2017. Additionally, owing to the near real time (NRT) data downlinking capabilities of the ISS, CATS provides expedited NRT data products within 6 hours of observation time. This capability provides a unique opportunity for supporting field campaigns and for developing data assimilation techniques to improve simulated cloud and aerosol vertical distributions in models. We additionally present preliminary work toward assimilating CATS observations into the NASA Goddard Earth Observing System version 5 (GEOS-5) global atmospheric model and data assimilation system.
Real-Time Sensor Validation, Signal Reconstruction, and Feature Detection for an RLV Propulsion Testbed
A real-time system for validating sensor health has been developed in support of the reusable launch vehicle program. This system was designed for use in a propulsion testbed as part of an overall effort to improve the safety, diagnostic capability, and cost of operation of the testbed. The sensor validation system was designed and developed at the NASA Lewis Research Center and integrated into a propulsion checkout and control system as part of an industry-NASA partnership, led by Rockwell International for the Marshall Space Flight Center. The system includes modules for sensor validation, signal reconstruction, and feature detection and was designed to maximize portability to other applications. Review of test data from initial integration testing verified real-time operation and showed the system to perform correctly on both hard and soft sensor failure test cases. This paper discusses the design of the sensor validation and supporting modules developed at LeRC and reviews results obtained from initial test cases.
Ion and electron energy dispersion features detected by ISEE 1
An energy dispersion feature observed by the ISEE 1 medium energy particle instrument is reported. This feature is characterized by enhancements in both electron and ion fluxes which occur in the 20-200 keV energy range and at radial distances from 5 to 13.5 earth radii. There is a strong energy dependence in the onset times of the enhancements. The time period between the first appearance of an enhancement at high energy to its eventual disappearance in the 20 keV channel can be as much as an hour. The duration of the enhancement in a given energy channel decreases with increasing energy. The species and pitch angle of particles in which the feature is most marked depend on local time. It is shown that the feature is a new manifestation of an old phenomenon, the drift dispersion that frequently produces signatures in particle data from geosynchronous satellites.
Meta-Analysis of Electron Cyclotron Resonance Absorption Features Detected in High-Mass X-Ray Binaries
Using recent compilations of detailed X-ray observations and spectral models of exceptional quality, we record the electron cyclotron resonance absorption (ECRA) features that have been detected in 45 pulsating high-mass X-ray binaries (HMXBs) and ultraluminous X-ray (ULX) sources harboring neutron stars, although seven of these detections are still questionable and another 21 are single and/or not independently confirmed. From the comprehensive catalogs of Jaisawal & Naik and Staubert et al. and from several additional recent observations, we produce two lists of HMXB ECRA sources: a list of 17 sources in which multiple ECRA lines or single very low-energy lines are seen, in which we can reasonably assume that the lowest energy reveals the fundamental cyclotron level for each source; and a "contaminated" list of 38 sources including the 21 detections of single ECRA lines that may (not) be higher-level harmonics. Both lists confirm a previous result that we have obtained independently by modeling the propeller lines of Magellanic HMXB pulsars: the surface dipolar magnetic fields B * of HMXB neutron stars are segregated around five distinct values with B * = 0.28 ± 0.08, 0.55 ± 0.11, 1.3 ± 0.37, 3.0 ± 0.68 and 7.9 ± 3.1, in units of TG. However, an explanation of this phenomenon is currently lacking. We have found no correlation between these B * values and the corresponding observed spin periods, spin period derivatives, orbital periods, maximum X-ray luminosities, neutron star masses or companion star masses.
The Use of Linear Feature Detection to Investigate Thematic Mapper Data Performance and Processing
Geometric and radiometric characteristics of Thematic Mapper data are investigated through analysis of linear features in the data. A linear feature is defined as two close, parallel and opposite edges. Examples in remotely sensed data are such features as rivers and roads. The geometric and radiometric precision TM data is sufficient to allow accurate measurement of linear feature widths. Results also confirm a 28.5m ground IFOV as specified prior to launch. The increase dimensionality of the TM data as compared with MSS data allows the possibility of independent verification of results by using data from several bands.
Patch-Based Convolutional Neural Networks for Multiple Microstructural Features Detection in FIB-SEM Micrographs of Irradiated Nuclear Fuel
Focused ion beam scanning electron microscopy (FIB-SEM) tomography has increasingly been utilized for acquiring three-dimensional (3D) microstructure features at the sub-micron scale in irradiated nuclear materials. This technique involves sequential ion beam slicing followed by electron beam imaging and compositional mapping using energy dispersive spectroscopy (EDS). Despite its growing use, several challenges persist. These include the time-intensive nature of data collection of EDS data, difficulties in distinguishing between various microstructures, and issues with image alignment. These challenges currently limit the broader application of FIB-SEM tomography in the field. To overcome these limitations, we propose using convolutional neural networks (CNNs) to automate microstructure identification in SEM images. Our study introduces a new framework for identifying microstructures in irradiated U-10Zr (wt. %) metallic fuel with limited annotated data. The framework includes the creation of a reliable annotated dataset with paired SEM and ground truth data from EDS maps, the applications of CNNs for microstructure identification, and the validation of model performance. Specifically, we employed the Segment Anything Model (SAM) to align SEM images with corresponding EDS maps and focused ion beam (FIB) tomography SEM data. We evaluate several models, including Patch-based U-Net, Attention U-Net, and Residual U-Net, finding that patch-based U-Net exhibits superior segmentation performance and consistency. This approach reduces reliance on EDS detectors and aids in accelerating nuclear material analysis process, highlighting the potential of advanced deep learning techniques to improve microstructural understanding in nuclear material. This is the first framework to integrate SAM and Patch-based CNN models for semantic segmentation of irradiated nuclear materials, with potential applicability to other tomography datasets.