SEARCH · Engineering Papers
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.
ss-rVAE classification can generalize from a small labeled data subset with weak orientational disorder to a larger unlabeled dataset with stronger disorder. We apply it to nanoparticle datasets to train a robust classifier and understand physical factors of data variation.
Crop residues, the portion of the crop left in the field after harvest, can be an important management factor in controlling soil erosion. Methods to quantify residue cover are needed that are rapid, accurate, and objective. Scenes with known amounts of crop residue were illuminated with long wave ultraviolet (UV) radiation and fluorescence images were recorded with an intensified video camera fitted with a 453 to 488 nm band pass filter. A light colored soil and a dark colored soil were used as background for the weathered soybean stems. Residue cover was determined by counting the proportion of the pixels in the image with fluorescence values greater than a threshold. Soil pixels had the lowest gray levels in the images. The values of the soybean residue pixels spanned nearly the full range of the 8-bit video data. Classification accuracies typically were within 3(absolute units) of measured cover values. Video imaging can provide an intuitive understanding of the fraction of the soil covered by residue.
Various methods are compared for carrying out land cover classifications of South America using multitemporal Advanced Very High Resolution Radiometer data. Fifty-two images of the normalized difference vegetation index (NDVI) from a 1-year period are used to generate multitemporal data sets. Three main approaches to land cover classification are considered, namely the use of the principal components transformed images, the use of a characteristic curves procedure based on NDVI values plotted against time, and finally application of the maximum likelihood rule to multitemporal data sets. Comparison of results from training sites indicates that the last approach yields the most accurate results. Despite the reliance on training site figures for performance assessment, the results are nevertheless extremely encouraging, with accuracies for several cover types exceeding 90 per cent.
Results are presented of an investigation to determine the degree to which digitally processed Landsat TM imagery can be used to discriminate among vegetated lava flows of different ages in the Menengai Caldera, Kenya. A selective series of five images, consisting of a color-coded Landsat 5 classification and four color composites, are compared with geologic maps. The most recent of more than 70 postcaldera flows within the caldera are trachytes, which are variably covered by shrubs and subsidiary grasses. Soil development evolves as a function of time, and as such supports a changing plant community. Progressively older flows exhibit the increasing dominance of grasses over bushes. The Landsat images correlated well with geologic maps, but the two mapped age classes could be further subdivided on the basis of different vegetation communities. It is concluded that field maps can be modified, and in some cases corrected by use of such imagery, and that digitally enhanced Landsat imagery can be a useful aid to field mapping in similar terrains.
Cloud-type classification based on multispectral satellite imagery data has been widely researched and demonstrated to be useful for distinguishing a variety of classes using a wide range of methods. The research described here is a comparison of the classifier output from two very different algorithms applied to Geostationary Operational Environmental Satellite (GOES) data over the course of one year. The first algorithm employs spectral channel thresholding and additional physically based tests. The second algorithm was developed through a supervised learning method with characteristic features of expertly labeled image samples used as training data for a 1-nearest-neighbor classification. The latter's ability to identify classes is also based in physics, but those relationships are embedded implicitly within the algorithm. A pixel-to-pixel comparison analysis was done for hourly daytime scenes within a region in the northeastern Pacific Ocean. Considerable agreement was found in this analysis, with many of the mismatches or disagreements providing insight to the strengths and limitations of each classifier. Depending upon user needs, a rule-based or other postprocessing system that combines the output from the two algorithms could provide the most reliable cloud-type classification.
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.
Human beings judge patterns (such as images) by complex mental processes, some of which may not be known, while computing machines extract features. By representing the human judgements with simple measurements and reducing them and the machine extracted features to a common metric space and fitting them by regression, the judgements of human experts rendered on a sample of patterns may be imposed on a pattern population to provide automatic classification.
Human beings judge patterns (such as images) by complex mental processes, some of which may not be known, while computing machines extract features. By representing the human judgements with simple measurements and reducing them and the machine extracted features to a common metric space and fitting them by regression, the judgements of human experts rendered on a sample of patterns may be imposed on a pattern population to provide automatic classification.
HPC is designed for large-scale simulations using monolithic codes of tightly coupled processes highly optimized to deliver decreased time to solution. Medical image processing is not a traditional field of HPC. Similar to AI applications, medical image processing parses large datasets, typically multiple times, to support a variety of studies for classification, diagnosis or monitoring purposes. The convergence of AI, HPC and Big Data encouraged more fields using image processing to transition to HPC. However, not all applications benefit from the same optimizations. In this paper we focus on high throughput medical image processing applications that analyze a huge dataset of small MRI images and that require HPC systems to decrease the time of parsing the entire dataset and not individual MRIs. We show in this research the performance of running SLANT, an image processing application for a whole brain segmentation, on large-scale systems and highlight performance limitations. We present optimizations prioritizing throughput that exhibit a 3.5x speed-up on the Summit Supercomputer that can be used as a baseline for building a high-throughput execution framework for other HPC systems.
Although a variety of techniques are available today for gray-scale image compression, a complete evaluation of these techniques cannot be made as there is no single reliable objective criterion for measuring the error in compressed images. The traditional subjective criteria are burdensome, and usually inaccurate or inconsistent. On the other hand, being the most common objective criterion, the mean square error (MSE) does not have a good correlation with the viewer's response. It is now understood that in order to have a reliable quality measure, a representative model of the complex human visual system is required. In this paper, we survey and give a classification of the criteria for the evaluation of monochrome image quality.
ABSTRACT We present in this paper one of the largest galaxy morphological classification catalogues to date, including over 20 million galaxies, using the Dark Energy Survey (DES) Year 3 data based on convolutional neural networks (CNNs). Monochromatic i-band DES images with linear, logarithmic, and gradient scales, matched with debiased visual classifications from the Galaxy Zoo 1 (GZ1) catalogue, are used to train our CNN models. With a training set including bright galaxies (16 ≤ i < 18) at low redshift (z < 0.25), we furthermore investigate the limit of the accuracy of our predictions applied to galaxies at fainter magnitude and at higher redshifts. Our final catalogue covers magnitudes 16 ≤ i < 21, and redshifts z < 1.0, and provides predicted probabilities to two galaxy types – ellipticals and spirals (disc galaxies). Our CNN classifications reveal an accuracy of over 99 per cent for bright galaxies when comparing with the GZ1 classifications (i < 18). For fainter galaxies, the visual classification carried out by three of the co-authors shows that the CNN classifier correctly categorizes discy galaxies with rounder and blurred features, which humans often incorrectly visually classify as ellipticals. As a part of the validation, we carry out one of the largest examinations of non-parametric methods, including ∼100 ,000 galaxies with the same coverage of magnitude and redshift as the training set from our catalogue. We find that the Gini coefficient is the best single parameter discriminator between ellipticals and spirals for this data set.
Rapid progress in aberration corrected electron microscopy necessitates development of robust methods for the identification of phases, ferroic variants, and other pertinent aspects of materials structure from imaging data. While unsupervised methods for clustering and classification are widely used for these tasks, their performance can be sensitive to hyperparameter selection in the analysis workflow. In this study, the effects of descriptors and hyperparameters are explored on the capability of unsupervised ML methods to distill local structural information, exemplified by the discovery of polarization and lattice distortion in Sm − dopped BiFeO 3 (BFO) thin films. It is demonstrated that a reward-driven approach can be used to optimize these key hyperparameters across the full workflow, where rewards are designed to reflect domain wall continuity and straightness, ensuring that the analysis aligns with the material's physical behavior. This approach allows the discovery of local descriptors that are best aligned with the specific physical behavior, providing insight into the fundamental physics of materials. The reward driven workflow is further extended to disentangle structural factors of variation via an optimized variational autoencoder (VAE). Lastly, the importance of well-defined rewards is explored as a quantifiable measure of the success of the workflow.
Multispectral imaging (MSI) plays a critical role in material classification, environmental monitoring, and remote sensing. However, MSI sensors typically have wavelength-dependent resolution, which limits downstream analysis. MSI super-resolution (MSI-SR) methods address this limitation by reconstructing all bands at a common high spatial resolution. Existing methods can achieve high reconstruction quality but often rely on spatially-coupled optimization or large learning-based models, leading to significant computational cost and limiting their use in large-scale or time-critical settings. In this paper, we introduce ResSR, a computationally efficient, model-based MSI-SR method that achieves high-quality reconstruction without supervised training or spatially-coupled optimization. Notably, ResSR decouples spectral and spatial processing into two sequential steps. ResSR first computes a spectrally-informed high-resolution estimate of the MSI using singular value decomposition together with a spatially-decoupled approximate forward model. It then applies a residual correction step to restore low-frequency spatial consistency while preserving high-frequency detail recovered by the spectral reconstruction. ResSR achieves comparable or improved reconstruction quality relative to existing MSI-SR methods while being
We present visual classifications of merger-induced tidal disturbances in 143 M* ~ 10 11 M ⊙ post-starburst galaxies at z ~ 0.7 identified in the $SQuIGG\vec{L}E$ Sample. This sample spectroscopically selects galaxies from the Sloan Digital Sky Survey that have stopped their primary epoch of star formation within the past ~500 Myr. Visual classifications are performed on Hyper Suprime-Cam imaging. We compare to a control sample of mass- and redshift-matched star-forming and quiescent galaxies from the Large Early Galaxy Census and find that post-starburst galaxies are more likely to be classified as disturbed than either category. This corresponds to a factor of $3.6$$^{+2.9}_{-1.3}$ times the disturbance rate of older quiescent galaxies and $2.1$$^{+1.9}_{-.73}$ times the disturbance rate of star-forming galaxies. Assuming tidal features persist for ≲500 Myr, this suggests merging is coincident with quenching in a significant fraction of these post-starbursts. Galaxies with tidal disturbances are younger on average than undisturbed post-starburst galaxies in our sample, suggesting tidal features from a major merger may have faded over time. This may be exacerbated by the fact that, on average, the undisturbed subset is fainter, rendering low-surface-brightness tidal features harder to identify. However, the presence of 10 young (≲150 Myr since quenching) undisturbed galaxies suggests that major mergers are not the only fast physical mechanism that shut down the primary epoch of star formation in massive galaxies at intermediate redshift.
Radar is considered as a means for delimiting associations of physical and cultural phenomena by outlining image variation in tone, texture, pattern, and shape. Classification and description of landscape types are discussed. The imagery used for the study covers a portion of the Asheville Basin in North Carolina, extending from Hot Springs to an area southeast of Hendersonville. Eleven regions are discussed.
The author has identified the following significant results. Improvement in the land use classification accuracy of ERTS-1 MSS multi-images over Kansas can be made using two distances between neighboring grey tone N-tuples instead of one distance. Much more information is contained texturally than spectrally on the Kansas image. Ground truth measurements indicate that reflectance ratios of the 545 and 655 nm wavebands provide an index of plant development and possibly physiological stress. Preliminary analysis of MSS 4 and 5 channels substantiate the ground truth interpretation. Results of the land use mapping experiment indicate that ERTS-1 imagery has major potential in regionalization. The ways in which land is utilized within these regions may then be studied more effectively than if no adequate regionalization is available. A model for estimating wheat yield per acre has been applied to acreage estimates derived from ERTS-1 imagery to project the 1973 wheat yields for a ten county area in southwest Kansas. The results are within 3% of the preharvest estimates for the same area prepared by the USDA. Visual identification of winter wheat is readily achieved by using a temporal sequence of images. Identification can be improve by stratifying the project area into subregions having more or less homogeneous agricultural practices and crop mixes.
The view angle effect in LANDSAT 2 imagery was investigated. The LANDSAT multispectral scanner scans over a range of view angles of -5.78 to 5.78 degrees. The view angle effect, which is caused by differing view angles, could be studied by comparing data collected at different view angles over a fixed location at a fixed time. Since such LANDSAT data is not available, consecutive day acquisition data were used as a substitute: they were collected over the same geographical location, acquired 24 hours apart, with a view angle change of 7 to 8 degrees at a latitude of 35 to 45 degrees. It is shown that there is approximately a 5% reduction in the average sensor response on the second-day acquisitions as compared with the first-day acquisitions, and that the view angle effect differs field to field and crop to crop. On false infrared color pictures the view angle effect causes changes primarily in brightness and to a lesser degree in color (hue and saturation). An implication is that caution must be taken when images with different view angles are combined for classification and a signature extension technique needs to take the view angle effect into account.
The accuracy of land-cover classifications based on Landsat-4 TM and MSS images (obtained in August 1982) and airborne TMS images (obtained in September 1981) of the New Martinsville, West Virginia area is evaluated by comparison with ground-truth data. TM, TMS, and MSS are found to have overall mapping accuracies 80.1, 78.5, and 75.6 percent; agriculture/grass accuracies 62.0, 29.7, and 46.6 percent; and developed-area accuracies 67.2, 77.8, and 59.4 percent, respectively.