Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Overhead Image”

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 37 records · Page 2

Supporting data for "Frozen flume experiments indicate rapid permafrost riverbank erosion depends on bank roughness"

This dataset contains supporting files detailing five frozen flume experiments conducted at the Caltech Earth Surface Dynamics Laboratory to investigate rates of ablation-limited permafrost riverbank erosion under controlled conditions. Water flowed past a bank of saturated, frozen sand and ice and gradually eroded the bank by thawing pore ice and immediately entraining sand and washing it downstream. Experiments were scaled for flow hydraulics and heat transfer allowing comparisons between our results and natural permafrost riverbanks. For each experiment, we measured the initial and final sand bank topography using a Keyence laser scanner, water surface slope at 3-min intervals throughout the experiment using a Massa sonar scanner, bank erosion using 10-sec overhead timelapse imagery taken by an overhead camera, water and bank temperature using thermistors frozen into the sand bank and sampling at 2 Hz, and water discharge using an in-line flow meter. We include calibration data for the carriage (engineered by the Saint Anthony Falls Laboratory) used to make sonar and laser topography measurements. We also include calibration data for temperature sensors, water discharge measurements, and images of a regular grid placed in the flume to align overhead camera images with the carriage datum. Grain size analysis for the channel bed (gravel) was produced using a pebble count and bank sand was measured using a Camsizer X2. In addition to the five frozen experiments, we include sonar scans of water surface slope and Keyence scans of bed and bank topography for calibration experiments ran with an immobile gravel bank and bed.

54 ENVIRONMENTAL SCIENCES↗

Overhead Detection of Underground Nuclear Explosions by Multi-Spectral and Infrared Imaging

The Comprehensive Test Ban Treaty (CTBT) permits Multi-Spectral and InfraRed Imaging (MSIR) to be performed as part of an On-Site Inspection (OSI) for the purpose of reducing the search area for the location of a possible underground nuclear explosion (UNE). Dedicated airborne MSIR measurements have not been made in conjunction with historical or recent UNE’s, so satellite data has been used to determine if there are MSIR observables associated with recent UNE’s. In this work MSIR data from commercial satellites has been used to show that there are detectable surface observables which can be used to greatly reduce the search area for the location of the UNE. This has been demonstrated using Landsat data of the Indian, Pakistani, and North Korean UNE’s in the last 12 years, and with GeoEye-1 data for the North Korean tests. The techniques used typically identify a region of interest less than 1 km2 in size (compared to the nominal 1,000 km2 search area), and the few false positives have been resolvable as such by using visible imagery. The results of this study show that MSIR data from satellites can be used to help prepare the inspection team for an OSI.

58 GEOSCIENCES↗

DINO: Divergent node cloning for sustained redundancy in HPC

Complexity and scale of next generation HPC systems pose significant challenges in fault resilience methods such that contemporary checkpoint/restart (C/R) methods that address fail-stop behavior may be insufficient. Redundant computing has been proposed as an alternative at extreme scale. Triple redundancy has an advantage over C/R in that it can also detect silent data corruption (SDC) and then correct results via voting. However, current redundant computing approaches do not repair failed or corrupted replicas. Consequently, SDCs can no longer be detected after a replica failure since the system has been degraded to dual redundancy without voting capability. Hence, a job may have to be aborted if voting uncovers mismatching results between the remaining two replicas. And while replicas are logically equivalent, they may have divergent runtime states during job execution, which presents a challenge to simply creating new replicas dynamically. This problem is addressed by, DIvergent NOde cloning (DINO), a redundant execution environment that quickly recovers from hard failures. DINO consists of a novel node cloning service integrated into the MPI runtime system that solves the problem of consolidating divergent states among replicas on-the-fly. With DINO, after degradation to dual redundancy, a good replica can be quickly cloned so that triple redundancy is restored. We present experimental results over 9 NAS Parallel Benchmarks (NPB), Sweep3D and LULESH. Results confirm the applicability of the approach and the correctness of the recovery process and indicate that DINO can recover from failures nearly instantly. The cloning overhead depends on the process image size that needs to be transferred between source and destination of the clone operation and varies between 5.60 to 90.48 s. Here, simulation results with our model show that dual redundancy with DINO recovery always outperforms 2x and surpasses 3x redundancy on up to 1 million nodes. To the best of our knowledge, the design and implementation for repairing failed replicas in redundant MPI computing is unprecedented.

97 MATHEMATICS AND COMPUTING↗

Estimating vehicle fuel economy from overhead camera imagery and application for traffic control

In this work, we explore the ability to estimate vehicle fuel consumption using imagery from overhead fisheye lens cameras deployed as traffic sensors. We utilize this information to simulate vision-based control of a traffic intersection, with a goal of improving fuel economy with minimal impact to mobility. We introduce the ORNL Overhead Vehicle Dataset (OOVD), consisting of a data set of paired, labeled vehicle images from a ground-based camera and an overhead fisheye lens traffic camera. The data set includes segmentation masks based on Gaussian mixture models for vehicle detection. We show the dataset utility through three applications: the estimate of fuel consumption based on segmentation bounding boxes, vehicle discrimination for those vehicles with largest bounding boxes, and a fine-grained classification on a limited number of vehicle makes and models using a pre-trained set of convolutional neural network models. We compare these results with estimates based on a large open-source data set based on web-scraped imagery. Finally, we show the utility of the approach using reinforcement learning in a traffic simulator using the open source Simulation of Urban Mobility (SUMO) package. Our results show the feasibility of the approach for controlling traffic lights for better fuel efficiency based solely on visual vehicle estimates from commercial, fisheye lens cameras.

Karnowski, Thomas↗

Optimized Hybrid Megavoltage-Kilovoltage Imaging Protocol for Volumetric Prostate Arc Therapy

Purpose: To develop a real-time prostate position monitoring technique for modern arc radiotherapy through novel use of cine-megavoltage (MV) imaging, together with as-needed kilovoltage (kV) imaging. Methods and Materials: We divided the task of monitoring the intrafraction prostate motion into two steps for rotational deliveries: to detect potential target motion beyond a predefined threshold using MV images from different viewing angles by taking advantage of gantry rotation during arc therapy and to verify the displacement and determine whether intervention is needed using fiducial/tumor position information acquired from combined MV-kV imaging (by turning on the kV imager). A Varian Trilogy linear accelerator with an onboard kV imager was used to examine selected typical trajectories using a four-dimensional motion phantom. The performance of the algorithm was evaluated using phantom measurements and computer simulation for 536 Calypso-measured tracks from 17 patients. Results: Fiducial displacement relative to the MV beam was limited to within a range of 3 mm 99.9% of the time with <1 mm accuracy. On average, only {approx}0.5 intervention per arc delivery was needed to achieve this level of accuracy. Compared with other fluoroscopy-based tracking techniques, kV use was significantly reduced to an average of <15 times per arc delivery. Conclusion: By focusing the attention on detecting predefined abnormal motion (i.e., 'failure' detection) and using the inherent mechanism of gantry rotation during arc radiotherapy, the current approach provides high confidence regarding the prostate position in real time without the unwanted overhead of continuous or periodic kV imaging.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Combining batch execution and leasing using virtual machines.

As cluster computers are used for a wider range of applications, we encounter the need to deliver resources at particular times, to meet particular deadlines, and/or at the same time as other resources are provided elsewhere. To address such requirements, we describe a scheduling approach in which users request resource leases, where leases can request either as-soon-as-possible ('best-effort') or reservation start times. We present the design of a lease management architecture, Haizea, that implements leases as virtual machines (VMs), leveraging their ability to suspend, migrate, and resume computations and to provide leased resources with customized application environments. We discuss methods to minimize the overhead introduced by having to deploy VM images before the start of a lease. We also present the results of simulation studies that compare alternative approaches. Using workloads with various mixes of best-effort and advance reservation requests, we compare the performance of our VM-based approach with that of non-VM-based schedulers. We find that a VM-based approach can provide better performance (measured in terms of both total execution time and average delay incurred by best-effort requests) than a scheduler that does not support task pre-emption, and only slightly worse performance than a scheduler that does support task pre-emption. We also compare the impact of different VM image popularity distributions and VM image caching strategies on performance. These results emphasize the importance of VM image caching for the workloads studied and quantify the sensitivity of scheduling performance to VM image popularity distribution.

99 GENERAL AND MISCELLANEOUS↗

Region templates: Data representation and management for high-throughput image analysis

Compartment fires exhibit unique features associated with smoke accumulation and restricted air ventilation. The objective of the present study is to evaluate the ability of CFD-based fire models to simulate these features and in particular the effects of flame extinction that results from under-ventilated fire conditions. The study is performed using FireFOAM; FireFOAM is an advanced fire modeling software developed by FM Global and is based on a general-purpose open-source software called OpenFOAM. A new flame extinction model based on the concept of a critical value of the flame Damköhler number is incorporated into FireFOAM. The performance of the extinction model is evaluated via comparisons with a previously developed experimental database corresponding to a reduced-scale, heptane-fueled, compartment fire configuration. The numerical simulations also include a description of fuel evaporation driven by the computed gas-to-liquid heat feedback. Comparisons between experimental data and numerical results provide a suitable test bed to evaluate the ability of CFD-based fire models to describe the transition from over- to under-ventilated fire conditions, as well as the transition from extinction-free conditions to conditions in which the flame experiences partial or total quenching.

Energy & Fuels↗

Event-to-Video Conversion for Overhead Object Detection

Collecting overhead imagery using an event camera is desirable due to the energy efficiency of the image sensor compared to standard cameras. However, event cameras complicate downstream image processing, especially for complex tasks such as object detection. In this paper, we investigate the viability of event streams for overhead object detection. We demonstrate that across a number of standard modeling approaches, there is a significant gap in performance between dense event representations and corresponding RGB frames. We establish that this gap is, in part, due to a lack of overlap between the event representations and the pre-training data that the object detectors were initially trained on through a number of experiments. Then, apply an off-the-shelf event-to-video conversion tool that converts event streams into gray-scale video to close this gap. We demonstrate that this approach results in a large performance increase, outperforming even event-specific object detection techniques on our overhead target task. These results suggest that better aligning event representations with existing large pre-trained models may result in greater short-term performance gains compared to end-to-end event-specific architectural improvements.

machine learning (ML), computer vision, Neuromorph↗

All-optical image denoising using a diffractive visual processor

Abstract Image denoising, one of the essential inverse problems, targets to remove noise/artifacts from input images. In general, digital image denoising algorithms, executed on computers, present latency due to several iterations implemented in, e.g., graphics processing units (GPUs). While deep learning-enabled methods can operate non-iteratively, they also introduce latency and impose a significant computational burden, leading to increased power consumption. Here, we introduce an analog diffractive image denoiser to all-optically and non-iteratively clean various forms of noise and artifacts from input images – implemented at the speed of light propagation within a thin diffractive visual processor that axially spans <250 × λ, where λ is the wavelength of light. This all-optical image denoiser comprises passive transmissive layers optimized using deep learning to physically scatter the optical modes that represent various noise features, causing them to miss the output image Field-of-View (FoV) while retaining the object features of interest. Our results show that these diffractive denoisers can efficiently remove salt and pepper noise and image rendering-related spatial artifacts from input phase or intensity images while achieving an output power efficiency of ~30–40%. We experimentally demonstrated the effectiveness of this analog denoiser architecture using a 3D-printed diffractive visual processor operating at the terahertz spectrum. Owing to their speed, power-efficiency, and minimal computational overhead, all-optical diffractive denoisers can be transformative for various image display and projection systems, including, e.g., holographic displays.

36 MATERIALS SCIENCE↗

REMODEL: Rethinking Deep CNN Models to Detect and Count on a NeuroSynaptic System

In this work, we perform analysis of detection and counting of cars using a low-power IBM TrueNorth Neurosynaptic System. For our evaluation we looked at a publicly-available dataset that has overhead imagery of cars with context present in the image. The trained neural network for image analysis was deployed on the NS16e system using IBM's EEDN training framework. Through multiple experiments we identify the architectural bottlenecks present in TrueNorth system that does not let us deploy large neural network structures. Following these experiments we propose changes to CNN model to circumvent these architectural bottlenecks. The results of these evaluations have been compared with caffe-based implementations of standard neural networks that were deployed on a Titan-X GPU. Results showed that TrueNorth can detect cars from the dataset with 97.60% accuracy and can be used to accurately count the number of cars in the image with 69.04% accuracy. The car detection accuracy and car count (–/+ 2 error margin) accuracy are comparable to high-precision neural networks like AlexNet, GoogLeNet, and ResCeption, but show a manifold improvement in power consumption.

97 MATHEMATICS AND COMPUTING↗

Conditional Experts for Improved Building Damage Assessment Across Satellite Imagery View Angles

Rapid building damage assessment (BDA) is vital in guiding disaster response missions and estimating population distribution across impacted areas. While commercial satellite imagery providers have enabled near-daily monitoring of the Earth, near-realtime assessment of disaster scenarios frequently requires analysis of off-nadir imagery, as satellites are often far from impacted areas for at-nadir post-event imaging to occur Such scenarios are, however, underrepresented in existing BDA datasets and methodologies. With this motivation, we investigate generalization capabilities of current BDA practices across overhead view-angles and strategies for their improvement. Using a labeled dataset of images capturing conflict-related damages, we first train a baseline BDA architecture using imbalanced and balanced datasets with respect to view-angle. Then, we explore conditional convolutions parameterized on image features, image nadir, and their combination as a mechanism for conditioning on view-angles. Experiments demonstrate the limitations of current practice and the potential of conditional mechanisms to increase model robustness to view-angle variations.

Ambrozio Dias, Philipe↗

"Mtrack 1.0": A multi-vehicle, deterministic tracking algorithm

As overhead persistent surveillance sensors become more capable at acquiring wide-field image sequences for longer time-spans, the need to exploit such data becomes ever greater. A desirable use of such sensors is to be able to track all the observable vehicles while they persist in the imagery. Typically, this imagery is characterized by lower resolutions (e.g. >= 0.5 m/pixel ) and lower frame rates (e.g. {approx} few Hz). We describe our initial implementation of a multiple-vehicle tracking algorithm for overhead persistent surveillance imagery. Subsequent reports will then present results of applying this tracker to specific persistent surveillance imagery as well as algorithm improvements and extensions for dealing with wider-field imagery.

42 ENGINEERING↗

Ultra-scale vehicle tracking in low spatial-resolution and low frame-rate overhead video

Overhead persistent surveillance systems are becoming more capable at acquiring wide-field image sequences for long time-spans. The need to exploit this data is becoming ever greater. The ability to track a single vehicle of interest or to track all the observable vehicles, which may number in the thousands, over large, cluttered regions while they persist in the imagery either in real-time or quickly on-demand is very desirable. With this ability we can begin to answer a number of interesting questions such as, what are normal traffic patterns in a particular region or where did that truck come from? There are many challenges associated with processing this type of data, some of which we will address in the paper. Wide-field image sequences are very large with many thousands of pixels on a side and are characterized by lower resolutions (e.g. worse than 0.5 meters/pixel) and lower frame rates (e.g. a few Hz or less). The objects in the scenery can vary in size, density, and contrast with respect to the background. At the same time the background scenery provides a number of clutter sources both man-made and natural. We describe our current implementation of an ultrascale capable multiple-vehicle tracking algorithm for overhead persistent surveillance imagery as well as discuss the tracking and timing performance of the currently implemented algorithm which is aimed at utilizing grayscale electrooptical image sequences alone for the track segment generation.

42 ENGINEERING↗

Modeling Spatial Dependencies in High-Resolution Overhead Imagery

Human settlement regions with different physical and socio-economic attributes exhibit unique spatial characteristics that are often illustrated in high-resolution overhead imageries. For example- size, shape and spatial arrangements of man-made structures are key attributes that vary with respect to the socioeconomic profile of the neighborhood. Successfully modeling these attributes is crucial in developing advanced image understanding systems for interpreting complex aerial scenes. In this paper we present three different approaches to model the spatial context in the overhead imagery. First, we show that the frequency domain of the image can be used to model the spatial context [1]. The shape of the spectral energy contours characterize the scene context and can be exploited as global features. Secondly, we explore a discriminative framework based on the Conditional Random Fields (CRF) [2] to model the spatial context in the overhead imagery. The features derived from the edge orientation distribution calculated for a neighborhood and the associated class labels are used as input features to model the spatial context. Our third approach is based on grouping spatially connected pixels based on the low-level edge primitives to form support-regions [3]. The statistical parameters generated from the support-region feature distributions characterize different geospatial neighborhoods. We apply our approaches on high-resolution overhead imageries. We show that proposed approaches characterize the spatial context in overhead imageries.

99 GENERAL AND MISCELLANEOUS↗

Multicontrast x-ray computed tomography imaging using Talbot-Lau interferometry without phase stepping

Purpose: The purpose of this work is to demonstrate that multicontrast computed tomography (CT) imaging can be performed using a Talbot-Lau interferometer without phase stepping, thus allowing for an acquisition scheme like that used for standard absorption CT. Methods: Rather than using phase stepping to extract refraction, small-angle scattering (SAS), and absorption signals, the two gratings of a Talbot-Lau interferometer were rotated slightly to generate a moire pattern on the detector. A Fourier analysis of the moire pattern was performed to obtain separate projection images of each of the three contrast signals, all from the same single-shot of x-ray exposure. After the signals were extracted from the detector data for all view angles, image reconstruction was performed to obtain absorption, refraction, and SAS CT images. A physical phantom was scanned to validate the proposed data acquisition method. The results were compared with a phantom scan using the standard phase stepping approach. Results: The reconstruction of each contrast mechanism produced the expected results. Signal levels and contrasts match those obtained using the phase stepping technique. Conclusions: Absorption, refraction, and SAS CT imaging can be achieved using the Talbot-Lau interferometer without the additional overhead of long scan time and phase stepping.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Automated Registration of Vector Data to Overhead Imagery

The availability of open source, remote sensing-derived vector data has increased exponentially in recent years. Unfortunately, these vector data are rarely made available with the corresponding source images from which objects were extracted. As such, these derived datasets are commonly used in combination with target images which differ from the source. Satellite viewing geometry can cause objects, extracted from a source image, to appear shifted when overlaid with a target image. Whether for purely cartographic purposes, reusability of preexisting training labels, or any spatial analysis where spatial correspondence between vector and image are required, the following paragraphs outline a method to address this challenge through the automated registration of vector data to overhead imagery, including existing literature regarding this challenge, followed by a case study of Sioux Falls, South Dakota.

McKee, Jacob↗

AN INDUCTIVE MAPPING WITH CONVOLUTIONAL REPRESENTATIONS FOR HUMAN SETTLEMENT DETECTION: PRELIMINARY RESULTS

Undoubtedly, deep convolutional learning methods continue to improve performance in image-level classification for computer vision and remote sensing applications. However, the spatio-temporal nature of remote sensing imagery offers other interesting challenges in multiscale image understanding. Emerging opportunities include seeking fine-grained and neighborhood mapping with overhead imagery. Limitations due to lack of relevant scale ground-truth often mandates that these challenges are pursued as disjoint. We test this premise and explore representation spaces from a single deep convolutional network and their visualization to argue for a unified feature extraction framework. The objective is to utilize and re-purpose trained feature extractors without the need for network retraining on three remote sensing tasks i.e. superpixel mapping, pixel-level segmentation and semantic based image visualization. By leveraging the same convolutional feature extractors and viewing them as visual information extractors that encode different image representation spaces, we demonstrate surprising potential on multiscale experiments that incorporate edge-level details up to semantic-level informationWe test this premise and explore representation spaces from a single deep convolutional network and their visualization to argue for a novel unified feature extraction framework. The objective is to utilize and re-purpose trained feature extractors without the need for network retraining on three remote sensing tasks i.e. superpixel mapping, pixel-level segmentation and semantic based image visualization. By leveraging the same convolutional feature extractors and viewing them as visual information extractors that encode different image representation spaces, we demonstrate a preliminary inductive transfer learning potential on multiscale experiments that incorporate edge-level details up to semantic-level informationWe test this premise and explore representation spaces from a single deep convolutional network and their visualization to argue for a novel unified feature extraction framework. The objective is to utilize and re-purpose trained feature extractors without the need for network retraining on three remote sensing tasks i.e. superpixel mapping, pixel-level segmentation and semantic based image visualization. By leveraging the same convolutional feature extractors and viewing them as visual information extractors that encode different image representation spaces, we demonstrate a preliminary inductive transfer learning potential on multiscale experiments that incorporate edge-level details up to semantic-level information.

Lunga, Dalton↗

"Mtrack 2.0": An ultra-scale tracking algorithm for low-resolution overhead imagery

Overhead persistent surveillance systems are becoming more capable at acquiring wide-field image sequences for long time-spans. The need to exploit this data is becoming ever greater. The ability to track a single vehicle of interest or to track all the observable vehicles, which may number in the thousands, over large, cluttered regions while they persist in the imagery is very desirable. Typically, this imagery has many thousands of pixels on a side and is characterized by lower resolutions (e.g. {approx}0.5 meters/pixel to {approx}2.0 meters/pixel) and lower frame rates (e.g. {approx} sub-Hz to several Hz). We describe our ultra-scale capable implementation of a multiple-vehicle tracking algorithm for overhead persistent surveillance imagery. This work builds upon an earlier report, where now the algorithm has been modified for improved performance and has been substantially improved to handle much larger datasets in a much shorter time.

42 ENGINEERING↗