Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Image Classification”

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 757 records · Page 42
All sourcesNASA NTRSDOE OSTI
NASA NTRS1990

Topics in inference and decision-making with partial knowledge

Two essential elements needed in the process of inference and decision-making are prior probabilities and likelihood functions. When both of these components are known accurately and precisely, the Bayesian approach provides a consistent and coherent solution to the problems of inference and decision-making. In many situations, however, either one or both of the above components may not be known, or at least may not be known precisely. This problem of partial knowledge about prior probabilities and likelihood functions is addressed. There are at least two ways to cope with this lack of precise knowledge: robust methods, and interval-valued methods. First, ways of modeling imprecision and indeterminacies in prior probabilities and likelihood functions are examined; then how imprecision in the above components carries over to the posterior probabilities is examined. Finally, the problem of decision making with imprecise posterior probabilities and the consequences of such actions are addressed. Application areas where the above problems may occur are in statistical pattern recognition problems, for example, the problem of classification of high-dimensional multispectral remote sensing image data.

Safavian, S. Rasoul↗
NASA NTRS1992

Analyst variability in labeling of unsupervised classifications

Analyst variability in the labeling of unsupervised classifications is tested for Landsat 5 Thematic Mapper image products covering two test sites in southern California. The accuracy of results are tested using samples from a photo interpreted base map of the area. The significance of differences between analysts is indicated by comparing Kappa statistics derived from error matrices. Analyst variability is found to be statistically significant in most cases. Certain analysts provided consistently better results for a given study area or degree of training. This work demonstrates the potential influence of analyst bias on what would otherwise seem to be a fairly objective method and suggests that controls for this subjectivity should be factored into experimental designs.

Mcgwire, Kenneth C.↗
NASA NTRS1993

A review and analysis of neural networks for classification of remotely sensed multispectral imagery

A literature survey and analysis of the use of neural networks for the classification of remotely sensed multispectral imagery is presented. As part of a brief mathematical review, the backpropagation algorithm, which is the most common method of training multi-layer networks, is discussed with an emphasis on its application to pattern recognition. The analysis is divided into five aspects of neural network classification: (1) input data preprocessing, structure, and encoding; (2) output encoding and extraction of classes; (3) network architecture, (4) training algorithms; and (5) comparisons to conventional classifiers. The advantages of the neural network method over traditional classifiers are its non-parametric nature, arbitrary decision boundary capabilities, easy adaptation to different types of data and input structures, fuzzy output values that can enhance classification, and good generalization for use with multiple images. The disadvantages of the method are slow training time, inconsistent results due to random initial weights, and the requirement of obscure initialization values (e.g., learning rate and hidden layer size). Possible techniques for ameliorating these problems are discussed. It is concluded that, although the neural network method has several unique capabilities, it will become a useful tool in remote sensing only if it is made faster, more predictable, and easier to use.

Paola, Justin D.↗
NASA NTRS2004

Communications: Mosquito Habitats, Land Use, and Malaria Risk in Belize from Satellite Imagery

Satellite imagery of northern Belize is used to examine the distribution of land use and breeding habitats of the malaria vector the Anopheles mosquito. A land cover classification based on multispectral SPOT and multitemporal Radarsat images identified eleven land cover classes, including agricultural, forest, and marsh types. Two of the land cover types, Typha domingensis marsh and flooded forest, are Anopheles vestitipennis larval habitats, and one, Eleocharis spp. marsh, is the larval habitat for Anopheles albimanus. Geographic Information Systems (GIS) analyses of land cover demonstrate that the amount of Typha domingensis in a marsh is positively correlated with the amount of agricultural land in the adjacent upland, and negatively correlated with the amount of adjacent forest. This finding is consistent with the hypothesis that nutrient (phosphorus) runoff from agricultural lands is causing an expansion of Typha domingensis in northern Belize. Thus, land use induced expansion of Anopheles vestitipennis larval habitat is potentially increasing malaria risk in Belize, and in other regions where Anopheles vestitipennis is a major malaria vector.

Pope, Kevin↗
NASA NTRSDate not supplied

Assessing Stratospheric Aerosols Contamination Due to Space Activities

The life cycle of a spacecraft starts and ends in the atmosphere: it interacts with the atmosphere right after the launch and during the atmospheric reentry when it usually mostly ablates. Both phases induce emissions of gases and solid particles, providing a source of these components in the middle atmosphere. Little is known about the exact nature, composition and effects of these emissions on the atmosphere and climate, but their impact is expected to rise as more and more orbiting satellites are launched. Ever since the years 2000, the number of space rockets launched per year has increased by a factor 3 globally. At the same time, the number of satellites launched in orbit around the Earth per year has been multiplied by about 30. One may wonder whether changes in the anthropogenic material injected in the terrestrial stratosphere can be detected and what its influence may be. In order to study the cosmic dust particles arriving on Earth, the NASA Johnson Space Center (JSC) has been systematically collecting solid dust particles from the Earth’s stratosphere by aircraft equipped with dedicated particle collectors since 1981. So far, 25 catalogs have been published, covering campaigns of collection from 1981 to 2020, with a total of 5071 solid particles that have been preliminary characterized and curated. In this work, we use the preliminary classification of the dust particles. Based on SEM images and X-ray EDS composition the collected dust is separated into four groups: C (Cosmic), TCN (Terrestrial Contaminant Natural), TCA (Terrestrial Contaminant Artificial) and AOS (Aluminum Oxide Sphere). The AOS being mostly generated by solid rocket propellant, they also belong to the TCA class. Our analysis of the data published indicates that from 1980 to 2009 the cosmic dust particles typically represent on average 40% of the collection with TCA and TCN corresponding to about 30% each. In the recent years, the TCA fraction has doubled to about 60% of the collection (2010-2020). This increase in anthropogenic particles is likely due to the overall human space activity and its recent increase. We will present the properties of the solid stratospheric dust particles collected and their evolution with time. Future work will be dedicated to better classify the natural and anthropogenic particles collected and described in the existing databases. We will use numerical modelling to produce quantitative estimates of the injected mass, the lifetime of particles in the middle atmosphere (stratosphere) and the relative abundance of the anthropogenic particles with respect to the stratospheric background particle population.

J Lasue↗
NASA NTRS1977

Status report: Data management program algorithm evaluation activity at Marshall Space Flight Center

An algorithm evaluation activity was initiated to study the problems associated with image processing by assessing the independent and interdependent effects of registration, compression, and classification techniques on LANDSAT data for several discipline applications. The objective of the activity was to make recommendations on selected applicable image processing algorithms in terms of accuracy, cost, and timeliness or to propose alternative ways of processing the data. As a means of accomplishing this objective, an Image Coding Panel was established. The conduct of the algorithm evaluation is described.

Jayroe, R. R., Jr.↗
NASA NTRS1977

Application of LANDSAT data to wetland study and land use classification in west Tennessee

The Obion-Forked Deer River Basin in northwest Tennessee is confronted with several acute land use problems which result in excessive erosion, sedimentation, pollution, and hydrologic runoff. LANDSAT data was applied to determine land use of selected watershed areas within the basin, with special emphasis on determining wetland boundaries. Densitometric analysis was performed to allow numerical classification of objects observed in the imagery on the basis of measurements of optical densities. Multispectral analysis of the LANDSAT imagery provided the capability of altering the color of the image presentation in order to enhance desired relationships. Manual mapping and classification techniques were performed in order to indicate a level of accuracy of the LANDSAT data as compared with high and low altitude photography for land use classification.

Jones, N. L.↗
NASA NTRS1996

Image analysis by geostatistical and neural-network methods applications in glaciology

The applicability of neural network techniques, in the classification of ice surfaces and crevasse patterns, was analyzed. The observations of the Bering Glacier (Alaska) obtained from a surface survey and from the global positioning system (GPS) were used. A geographical information system was applied to test the usefulness of standard approaches. The information in the image needed to be reduced prior to the classification. The reduction was performed with a fast variogram algorithm sampling in three oblique directions. The resultant vectors provided the input for the neural network.

Herzfeld, Ute Christina↗
NASA NTRS1983

Shuttle Imaging Radar-A (SIR-A) data as a complement to Landsat Multispectral Scanner (MSS) data

Principal components analysis and supervised classifications were performed on two dates of Landsat multispectral scanner (MSS) data registered to one date of Shuttle Imaging Radar-A (SIR-A) data in a wheat-growing area of New South Wales, Australia. The purpose was to evaluate SIR-A data as a complement to Landsat MSS data in an agricultural environment. The SIR-A data was filtered using a 7 x 7 pixel moving window median filter. Principal components analysis indicated the SIR-A data were discriminating between trees and agricultural fields. Supervised classifications using wheat, pasture, trees, and idle classes resulted in increased accuracies for wheat and pasture and slightly decreased accuracies for trees and idle for the Landsat MSS/SIR-A registered data sets over the Landsat MSS alone. Overall classification accuracies were unchanged for one date and substantially increased for the other when the SIR-A data were added to the Landsat MSS data.

Henninger, D. L.↗
NASA NTRS2007

Automated Recognition of 3D Features in GPIR Images

A method of automated recognition of three-dimensional (3D) features in images generated by ground-penetrating imaging radar (GPIR) is undergoing development. GPIR 3D images can be analyzed to detect and identify such subsurface features as pipes and other utility conduits. Until now, much of the analysis of GPIR images has been performed manually by expert operators who must visually identify and track each feature. The present method is intended to satisfy a need for more efficient and accurate analysis by means of algorithms that can automatically identify and track subsurface features, with minimal supervision by human operators. In this method, data from multiple sources (for example, data on different features extracted by different algorithms) are fused together for identifying subsurface objects. The algorithms of this method can be classified in several different ways. In one classification, the algorithms fall into three classes: (1) image-processing algorithms, (2) feature- extraction algorithms, and (3) a multiaxis data-fusion/pattern-recognition algorithm that includes a combination of machine-learning, pattern-recognition, and object-linking algorithms. The image-processing class includes preprocessing algorithms for reducing noise and enhancing target features for pattern recognition. The feature-extraction algorithms operate on preprocessed data to extract such specific features in images as two-dimensional (2D) slices of a pipe. Then the multiaxis data-fusion/ pattern-recognition algorithm identifies, classifies, and reconstructs 3D objects from the extracted features. In this process, multiple 2D features extracted by use of different algorithms and representing views along different directions are used to identify and reconstruct 3D objects. In object linking, which is an essential part of this process, features identified in successive 2D slices and located within a threshold radius of identical features in adjacent slices are linked in a directed-graph data structure. Relative to past approaches, this multiaxis approach offers the advantages of more reliable detections, better discrimination of objects, and provision of redundant information, which can be helpful in filling gaps in feature recognition by one of the component algorithms. The image-processing class also includes postprocessing algorithms that enhance identified features to prepare them for further scrutiny by human analysts (see figure). Enhancement of images as a postprocessing step is a significant departure from traditional practice, in which enhancement of images is a preprocessing step.

Park, Han↗
DOE OSTI2022

AICCA: AI-Driven Cloud Classification Atlas

Clouds play an important role in the Earth’s energy budget, and their behavior is one of the largest uncertainties in future climate projections. Satellite observations should help in understanding cloud responses, but decades and petabytes of multispectral cloud imagery have to date received only limited use. This study describes a new analysis approach that reduces the dimensionality of satellite cloud observations by grouping them via a novel automated, unsupervised cloud classification technique based on a convolutional autoencoder, an artificial intelligence (AI) method good at identifying patterns in spatial data. Our technique combines a rotation-invariant autoencoder and hierarchical agglomerative clustering to generate cloud clusters that capture meaningful distinctions among cloud textures, using only raw multispectral imagery as input. Cloud classes are therefore defined based on spectral properties and spatial textures without reliance on location, time/season, derived physical properties, or pre-designated class definitions. We use this approach to generate a unique new cloud dataset, the AI-driven cloud classification atlas (AICCA), which clusters 22 years of ocean images from the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Aqua and Terra instruments—198 million patches, each roughly 100 km × 100 km (128 × 128 pixels)—into 42 AI-generated cloud classes, a number determined via a newly-developed stability protocol that we use to maximize richness of information while ensuring stable groupings of patches. AICCA thereby translates 801 TB of satellite images into 54.2 GB of class labels and cloud top and optical properties, a reduction by a factor of 15,000. The 42 AICCA classes produce meaningful spatio-temporal and physical distinctions and capture a greater variety of cloud types than do the nine International Satellite Cloud Climatology Project (ISCCP) categories—for example, multiple textures in the stratocumulus decks along the West coasts of North and South America. We conclude that our methodology has explanatory power, capturing regionally unique cloud classes and providing rich but tractable information for global analysis. AICCA delivers the information from multi-spectral images in a compact form, enables data-driven diagnosis of patterns of cloud organization, provides insight into cloud evolution on timescales of hours to decades, and helps democratize climate research by facilitating access to core data.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1981

Computer program documentation for the patch subsampling processor

The programs presented are intended to provide a way to extract a sample from a full-frame scene and summarize it in a useful way. The sample in each case was chosen to fill a 512-by-512 pixel (sample-by-line) image since this is the largest image that can be displayed on the Integrated Multivariant Data Analysis and Classification System. This sample size provides one megabyte of data for manipulation and storage and contains about 3% of the full-frame data. A patch image processor computes means for 256 32-by-32 pixel squares which constitute the 512-by-512 pixel image. Thus, 256 measurements are available for 8 vegetation indexes over a 100-mile square.

Nieves, M. J.↗
DOE OSTI2020

Deep transfer learning for star cluster classification: I. application to the PHANGS– HST survey

ABSTRACT We present the results of a proof-of-concept experiment that demonstrates that deep learning can successfully be used for production-scale classification of compact star clusters detected in Hubble Space Telescope(HST) ultraviolet-optical imaging of nearby spiral galaxies ($D\lesssim 20\, \textrm{Mpc}$) in the Physics at High Angular Resolution in Nearby GalaxieS (PHANGS)–HST survey. Given the relatively small nature of existing, human-labelled star cluster samples, we transfer the knowledge of state-of-the-art neural network models for real-object recognition to classify star clusters candidates into four morphological classes. We perform a series of experiments to determine the dependence of classification performance on neural network architecture (ResNet18 and VGG19-BN), training data sets curated by either a single expert or three astronomers, and the size of the images used for training. We find that the overall classification accuracies are not significantly affected by these choices. The networks are used to classify star cluster candidates in the PHANGS–HST galaxy NGC 1559, which was not included in the training samples. The resulting prediction accuracies are 70 per cent, 40 per cent, 40–50 per cent, and 50–70 per cent for class 1, 2, 3 star clusters, and class 4 non-clusters, respectively. This performance is competitive with consistency achieved in previously published human and automated quantitative classification of star cluster candidate samples (70–80 per cent, 40–50 per cent, 40–50 per cent, and 60–70 per cent). The methods introduced herein lay the foundations to automate classification for star clusters at scale, and exhibit the need to prepare a standardized data set of human-labelled star cluster classifications, agreed upon by a full range of experts in the field, to further improve the performance of the networks introduced in this study.

Wei, Wei↗
NASA NTRS2016

Geocam Space: Enhancing Handheld Digital Camera Imagery from the International Space Station for Research and Applications

Handheld astronaut photography of the Earth has been collected from the International Space Station (ISS) since 2000, making it the most temporally extensive remotely sensed dataset from this unique Low Earth orbital platform. Exclusive use of digital handheld cameras to perform Earth observations from the ISS began in 2004. Nadir viewing imagery is constrained by the inclined equatorial orbit of the ISS to between 51.6 degrees North and South latitude, however numerous oblique images of land surfaces above these latitudes are included in the dataset. While unmodified commercial off-the-shelf digital cameras provide only visible wavelength, three-band spectral information of limited quality current cameras used with long (400+ mm) lenses can obtain high quality spatial information approaching 2 meters/ground pixel resolution. The dataset is freely available online at the Gateway to Astronaut Photography of Earth site (http://eol.jsc.nasa.gov), and now comprises over 2 million images. Despite this extensive image catalog, use of the data for scientific research, disaster response, commercial applications and visualizations is minimal in comparison to other data collected from free-flying satellite platforms such as Landsat, Worldview, etc. This is due primarily to the lack of fully-georeferenced data products - while current digital cameras typically have integrated GPS, this does not function in the Low Earth Orbit environment. The Earth Science and Remote Sensing (ESRS) Unit at NASA Johnson Space Center provides training in Earth Science topics to ISS crews, performs daily operations and Earth observation target delivery to crews through the Crew Earth Observations (CEO) Facility on board ISS, and also catalogs digital handheld imagery acquired from orbit by manually adding descriptive metadata and determining an image geographic centerpoint using visual feature matching with other georeferenced data, e.g. Landsat, Google Earth, etc. The lack of full geolocation information native to the data makes it difficult to integrate astronaut photographs with other georeferenced data to facilitate quantitative analysis such as urban land cover/land use classification, change detection, or geologic mapping. The manual determination of image centerpoints is both time and labor-intensive, leading to delays in releasing geolocated and cataloged data to the public, such as the timely use of data for disaster response. The GeoCam Space project was funded by the ISS Program in 2015 to develop an on-orbit hardware and ground-based software system for increasing the efficiency of geolocating astronaut photographs from the ISS (Fig. 1). The Intelligent Robotics Group at NASA Ames Research Center leads the development of both the ground and on-orbit systems in collaboration with the ESRS Unit. The hardware component consists of modified smartphone elements including cameras, central processing unit, wireless Ethernet, and an inertial measurement unit (gyroscopes/accelerometers/magnetometers) reconfigured into a compact unit that attaches to the base of the current Nikon D4 camera - and its replacement, the Nikon D5 - and connects using the standard Nikon peripheral connector or USB port. This provides secondary, side and downward facing cameras perpendicular to the primary camera pointing direction. The secondary cameras observe calibration targets with known internal X, Y, and Z position affixed to the interior of the ISS to determine the camera pose corresponding to each image frame. This information is recorded by the GeoCam Space unit and indexed for correlation to the camera time recorded for each image frame. Data - image, EXIF header, and camera pose information - is transmitted to the ground software system (GeoRef) using the established Ku-band USOS downlink system. Following integration on the ground, the camera pose information provides an initial geolocation estimate for the individual film frame. This new capability represents a significant advance in geolocation from the manual feature-matching approach for both nadir and off-nadir viewing imagery. With the initial geolocation estimate, full georeferencing of an image is completed using the rapid tie-pointing interface in GeoRef, and the resulting data is added to the Gateway to Astronaut Photography of Earth online database in both Geotiff and Keyhole Markup Language (kml) formats. The integration of the GeoRef software component of Geocam Space into the CEO image cataloging workflow is complete, and disaster response imagery acquired by the ISS crew is now fully georeferenced as a standard data product. The on-orbit hardware component (GeoSens) is in final prototyping phase, and is on-schedule for launch to the ISS in late 2016. Installation and routine use of the Geocam Space system for handheld digital camera photography from the ISS is expected to significantly improve the usefulness of this unique dataset for a variety of public- and private-sector applications.

Stefanov, William L.↗
DOE OSTI2025

Hyperspectral imaging for real-time waste materials characterization and recovery using endmember extraction and abundance detection

Hyperspectral imaging, combined with advanced spectral unmixing techniques and artificial intelligence, offers a powerful solution for improving material identification and classification. Here, this study evaluates the effectiveness of the pixel purity index and the sequential maximum angle convex cone algorithms in extracting and validating spectral signatures from pure samples of paper components (cellulose and lignin) and plastic (polypropylene). Principal-component analysis showed that both algorithms captured nearly all relevant variance for the tested materials. Spectral signatures were compared using the spectral angle mapper, revealing high similarity in the short-wave infrared region and greater variability in the visible near-infrared range. The methodology was then applied to a disposable coffee cup to detect and quantify mixed materials, accurately estimating material abundance and object area with less than 1% error. This approach enhances material classification, supporting product verification, quality control, and automated sorting for sustainable waste management and resource recovery.

36 MATERIALS SCIENCE↗
NASA NTRS2016

Up Periscope! Designing a New Perceptual Metric for Imaging System Performance

Modern electronic imaging systems include optics, sensors, sampling, noise, processing, compression, transmission and display elements, and are viewed by the human eye. Many of these elements cannot be assessed by traditional imaging system metrics such as the MTF. More complex metrics such as NVTherm do address these elements, but do so largely through parametric adjustment of an MTF-like metric. The parameters are adjusted through subjective testing of human observers identifying specific targets in a set of standard images. We have designed a new metric that is based on a model of human visual pattern classification. In contrast to previous metrics, ours simulates the human observer identifying the standard targets. One application of this metric is to quantify performance of modern electronic periscope systems on submarines.

imaging↗
NASA NTRS1999

An Intercomparison Between Radar Reflectivity and the IR Cloud Classification Technique for the TOGA-COARE Area

Satellite infrared (IR) and visible (VIS) images from the Tropical Ocean Global Atmosphere - Coupled Ocean Atmosphere Response Experiment (TOGA-COARE) experiment are investigated through the use of Clustering Analysis. The clusters are obtained from the values of IR and VIS counts and the local variance for both channels. The clustering procedure is based on the standardized histogram of each variable obtained from 179 pairs of images. A new approach to classify high clouds using only IR and the clustering technique is proposed. This method allows the separation of the enhanced convection in two main classes: convective tops, more closely related to the most active core of the storm, and convective systems, which produce regions of merged, thick anvil clouds. The resulting classification of different portions of cloudiness is compared to the radar reflectivity field for intensive events. Convective Systems and Convective Tops are followed during their life cycle using the IR clustering method. The areal coverage of precipitation and features related to convective and stratiform rain is obtained from the radar for each stage of the evolving Mesoscale Convective Systems (MCS). In order to compare the IR clustering method with a simple threshold technique, two IR thresholds (Tir) were used to identify different portions of cloudiness, Tir=240K which roughly defines the extent of all cloudiness associated with the MCS, and Tir=220K which indicates the presence of deep convection. It is shown that the IR clustering technique can be used as a simple alternative to identify the actual portion of convective and stratiform rainfall.

Carvalho, L. M. V.↗
DOE OSTI2022

Photometric redshift estimation with convolutional neural networks and galaxy images: Case study of resolving biases in data-driven methods

Deep-learning models have been increasingly exploited in astrophysical studies, but these data-driven algorithms are prone to producing biased outputs that are detrimental for subsequent analyses. In this work, we investigate two main forms of biases: class-dependent residuals, and mode collapse. We do this in a case study, in which we estimate photometric redshift as a classification problem using convolutional neural networks (CNNs) trained with galaxy images and associated spectroscopic redshifts. We focus on point estimates and propose a set of consecutive steps for resolving the two biases based on CNN models, involving representation learning with multichannel outputs, balancing the training data, and leveraging soft labels. The residuals can be viewed as a function of spectroscopic redshift or photometric redshift, and the biases with respect to these two definitions are incompatible and should be treated individually. We suggest that a prerequisite for resolving biases in photometric space is resolving biases in spectroscopic space. Experiments show that our methods can better control biases than benchmark methods, and they are robust in various implementing and training conditions with high-quality data. Our methods hold promises for future cosmological surveys that require a good constraint of biases, and they may be applied to regression problems and other studies that make use of data-driven models. Nonetheless, the bias-variance tradeoff and the requirement of sufficient statistics suggest that we need better methods and optimized data usage strategies.

79 ASTRONOMY AND ASTROPHYSICS↗