Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “image recognition”

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 307 records · Page 17
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Evaluating Automated Face Identity-Masking Methods with Human Perception and a Deep Convolutional Neural Network

Face de-identification (or “masking”) algorithms have been developed in response to the prevalent use of video recordings in public places. Here, we evaluated the success of face identity masking for human perceivers and a deep convolutional neural network (DCNN). Eight de-identification algorithms were applied to videos of drivers’ faces, while they actively operated a motor vehicle. These masks were pre-selected to be applicable to low-quality video and to maintain coarse information about facial actions. Humans studied high-resolution images to learn driver identities and were tested on their recognition of active drivers in low-resolution videos. Faces in the videos were either unmasked or were masked by one of the eight algorithms. When participants were tested immediately after learning (Experiment 1), all masks reduced identification, with six of eight masks reducing identification to extremely poor performance. In a second experiment, two of the most effective masks were tested after a delay of 7 or 28 days. The delay did not further reduce identification of the masked faces. In all masked conditions, participants maintained stringent decision criteria, with low confidence in recognition, further indicating the effectiveness of the masks. Next, the DCNN performed an identity-matching task between high-resolution images and masked videos—a task analogous to that done by humans. The pattern of accuracy for the DCNN mirrored some, but not all, aspects of human performance, highlighting the need to test the effectiveness of identity masking for both humans and machines. The DCNN was also tested on its ability to match identity between masked and unmasked versions of the same video, based only on the face. DCNN performance for the eight masks offers insight into the nature of the information in faces that is coded in these networks.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1996

Active-Pixel Image Sensors With Programmable Resolution

Active-pixel image sensors with programmable resolution proposed for use in applications in which speed and efficiency of processing of image data enhanced by providing those data at varying resolutions. Such applications include modeling of biological vision, stereoscopic range-finding, recognition of patterns, tracking targets, and progressive transmission of compressed images. In target-tracking application, sensor initially forms low-resolution image from which area of interest identified, then sensor set at high resolution for examination of identified area. Outputs of contiguous pixels combined. Sensor of this type made to act as though it comprised fewer and larger pixels.

Kemeny, Sabrina E.↗
DOE OSTI2021

Neutrino event selection in the MicroBooNE liquid argon time projection chamber using Wire-Cell 3D imaging, clustering, and charge-light matching

An accurate and efficient event reconstruction is required to realize the full scientific capability of liquid argon time projection chambers (LArTPCs). The current and future neutrino experiments that rely on massive LArTPCs create a need for new ideas and reconstruction approaches. Wire-Cell, proposed in recent years, is a novel tomographic event reconstruction method for LArTPCs. The Wire-Cell 3D imaging approach capitalizes on charge, sparsity, time, and geometry information to reconstruct a topology-agnostic 3D image of the ionization electrons prior to pattern recognition. A second novel method, the many-to-many charge-light matching, then pairs the TPC charge activity to the detected scintillation light signal, thus enabling a powerful rejection of cosmic-ray muons in the MicroBooNE detector. A robust processing of the scintillation light signal and an appropriate clustering of the reconstructed 3D image are fundamental to this technique. In this paper, we describe the principles and algorithms of these techniques and their successful application in the MicroBooNE experiment. A quantitative evaluation of the performance of these techniques is presented. Using these techniques, a 95% efficient pre-selection of neutrino charged-current events is achieved with a 30-fold reduction of non-beam-coincident cosmic-ray muons, and about 80% of the selected neutrino charged-current events are reconstructed with at least 70% completeness and 80% purity.

3D imaging↗
NASA NTRS1981

Air Force research in optical processing

Optical and optical electronic hybrid processing especially in the application area of image processing are emphasized. Real time pattern recognition processors for such airborne missions as target recognition, tracking, and terminal guidance are studied.

Neff, J.↗
NASA NTRS1986

Coding for Electronic Mail

Scheme for coding facsimile messages promises to reduce data transmission requirements to one-tenth current level. Coding scheme paves way for true electronic mail in which handwritten, typed, or printed messages or diagrams sent virtually instantaneously - between buildings or between continents. Scheme, called Universal System for Efficient Electronic Mail (USEEM), uses unsupervised character recognition and adaptive noiseless coding of text. Image quality of resulting delivered messages improved over messages transmitted by conventional coding. Coding scheme compatible with direct-entry electronic mail as well as facsimile reproduction. Text transmitted in this scheme automatically translated to word-processor form.

Rice, R. F.↗
NASA NTRS1989

An improved architecture for video rate image transformations

Geometric image transformations are of interest to pattern recognition algorithms for their use in simplifying some aspects of the pattern recognition process. Examples include reducing sensitivity to rotation, scale, and perspective of the object being recognized. The NASA Programmable Remapper can perform a wide variety of geometric transforms at full video rate. An architecture is proposed that extends its abilities and alleviates many of the first version's shortcomings. The need for the improvements are discussed in the context of the initial Programmable Remapper and the benefits and limitations it has delivered. The implementation and capabilities of the proposed architecture are discussed.

Fisher, Timothy E.↗
NASA NTRS1994

Neural network wavelet technology: A frontier of automation

Neural networks are an outgrowth of interdisciplinary studies concerning the brain. These studies are guiding the field of Artificial Intelligence towards the, so-called, 6th Generation Computer. Enormous amounts of resources have been poured into R/D. Wavelet Transforms (WT) have replaced Fourier Transforms (FT) in Wideband Transient (WT) cases since the discovery of WT in 1985. The list of successful applications includes the following: earthquake prediction; radar identification; speech recognition; stock market forecasting; FBI finger print image compression; and telecommunication ISDN-data compression.

Szu, Harold↗
DOE OSTI2025

Quantum optical classifier with superexponential speedup

Abstract Classification is a central task in deep learning algorithms. Usually, images are first captured and then processed by a sequence of operations, of which the artificial neuron represents one of the fundamental units. This paradigm requires significant resources that scale (at least) linearly in the image resolution, both in terms of photons and computational operations. Here, we present a quantum optical pattern recognition method for binary classification tasks. It classifies objects without reconstructing their images, using the rate of two-photon coincidences at the output of a Hong-Ou-Mandel interferometer, where both the input and the classifier parameters are encoded into single-photon states. Our method exhibits the behaviour of a classical neuron of unit depth. Once trained, it shows a constant $${{\mathcal{O}}}(1)$$ O ( 1 ) complexity in the number of computational operations and photons required by a single classification. This is a superexponential advantage over a classical artificial neuron.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
NASA NTRS2012

Obtaining Accurate Change Detection Results from High-Resolution Satellite Sensors

Multi-date acquisitions of high-resolution imaging satellites (e.g. GeoEye and WorldView), can display local changes of current economic interest. However, their large data volume precludes effective manual analysis, requiring image co-registration followed by image-to-image change detection, preferably with minimal analyst attention. We have recently developed an automatic change detection procedure that minimizes false-positives. The processing steps include: (a) Conversion of both the pre- and post- images to reflectance values (this step is of critical importance when different sensors are involved); reflectance values can be either top-of-atmosphere units or have full aerosol optical depth calibration applied using bi-directional reflectance knowledge. (b) Panchromatic band image-to-image co-registration, using an orthorectified base reference image (e.g. Digital Orthophoto Quadrangle) and a digital elevation model; this step can be improved if a stereo-pair of images have been acquired on one of the image dates. (c) Pan-sharpening of the multispectral data to assure recognition of change objects at the highest resolution. (d) Characterization of multispectral data in the post-image ( i.e. the background) using unsupervised cluster analysis. (e) Band ratio selection in the post-image to separate surface materials of interest from the background. (f) Preparing a pre-to-post change image. (g) Identifying locations where change has occurred involving materials of interest.

multispectral↗
NASA NTRS1979

Quest for ultrahigh resolution in X-ray optics

A program of solar X-ray astronomy using grazing incidence optics has culminated in X-ray images of the corona having one arc second spatial resolution. These images have demonstrated that, in general, X-ray optics can be fabricated to their specifications and can provide the level of resolution for which they are designed. Several aspects of these programs relating to the performance of X-ray optics in regard to resolution, including the point response function, the variation of resolution with off-axis position and the recognition that nearly all solar X-ray images have been film limited, are discussed. By extending the experience gained on this and other programs it is clearly possible to design and fabricate X-ray optics with sub arc sec resolution. The performance required to meet the scientific objectives for the remainder of the century are discussed in relation to AXIO, an Advanced X-Ray Imaging Observatory for solar observations which is proposed for flight on the Space Shuttle. Several configurations of AXIO are described, each of which would be a major step in the quest for ultrahigh-resolution observations.

Davis, J. M.↗
DOE OSTI2024

Semi-automatic image annotation using 3D LiDAR projections and depth camera data

Efficient image annotation is necessary to utilize deep learning object recognition neural networks in nuclear safeguards, such as for the detection and localization of target objects like nuclear material containers (NMCs). This capability can help automate the inventory accounting of different types of NMCs within nuclear storage facilities. The conventional manual annotation process is labor-intensive and time-consuming, hindering the rapid deployment of deep learning models for NMC identifications. This paper introduces a novel semi-automatic method for annotating 2D images of nuclear material containers (NMCs) by combining 3D light detection and ranging (LiDAR) data with color and depth camera images collected from a handheld scan system. The annotation pipeline involves an operator manually marking new target objects on a LiDAR-generated map, and projecting these 3D locations to images, thereby automatically creating annotations from the projections. The semi-automatic approach significantly reduces manual efforts and the expertise in image annotation that is required to perform the task, allowing deep learning models to be trained on-site within a few hours. The paper compares the performance of models trained on datasets annotated through various methods, including semi-automatic, manual, and commercial annotation services. The evaluation demonstrates that the semi-automatic annotation method achieves comparable or superior results, with a mean average precision (mAP) above 0.9, showcasing its efficiency in training object recognition models. Additionally, the paper explores the application of the proposed method to instance segmentation, achieving promising results in detecting multiple types of NMCs in various formations.

98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL P↗
NASA NTRS1997

On the Implementation of a Land Cover Classification System for SAR Images Using Khoros

The Synthetic Aperture Radar (SAR) sensor is widely used to record data about the ground under all atmospheric conditions. The SAR acquired images have very good resolution which necessitates the development of a classification system that process the SAR images to extract useful information for different applications. In this work, a complete system for the land cover classification was designed and programmed using the Khoros, a data flow visual language environment, taking full advantages of the polymorphic data services that it provides. Image analysis was applied to SAR images to improve and automate the processes of recognition and classification of the different regions like mountains and lakes. Both unsupervised and supervised classification utilities were used. The unsupervised classification routines included the use of several Classification/Clustering algorithms like the K-means, ISO2, Weighted Minimum Distance, and the Localized Receptive Field (LRF) training/classifier. Different texture analysis approaches such as Invariant Moments, Fractal Dimension and Second Order statistics were implemented for supervised classification of the images. The results and conclusions for SAR image classification using the various unsupervised and supervised procedures are presented based on their accuracy and performance.

Medina Revera, Edwin J.↗
NASA NTRS1975

A study of payload specialist station monitor size constraints

Constraints on the CRT display size for the shuttle orbiter cabin are studied. The viewing requirements placed on these monitors were assumed to involve display of imaged scenes providing visual feedback during payload operations and display of alphanumeric characters. Data on target recognition/resolution, target recognition, and range rate detection by human observers were utilized to determine viewing requirements for imaged scenes. Field-of-view and acuity requirements for a variety of payload operations were obtained along with the necessary detection capability in terms of range-to-target size ratios. The monitor size necessary to meet the acuity requirements was established. An empirical test was conducted to determine required recognition sizes for displayed alphanumeric characters. The results of the test were used to determine the number of characters which could be simultaneously displayed based on the recognition size requirements using the proposed monitor size. A CRT display of 20 x 20 cm is recommended. A portion of the display area is used for displaying imaged scenes and the remaining display area is used for alphanumeric characters pertaining to the displayed scene. The entire display is used for the character alone mode.

Kirkpatrick, M., III↗
NASA NTRS1977

Hybrid optical/digital detector

Device for automatic-diffraction pattern recognition is designed around concentric-ring fiber-optic array. Coherent light passes through film-image transparency and lens producing Fourier transform of image. Transform is converted to digital data, the basis of diffraction-pattern recognition program performed by computer.

Shackelford, R. G.↗
NASA NTRS2003

Automated Coronal Loop Identification Using Digital Image Processing Techniques

The results of a master thesis project on a study of computer algorithms for automatic identification of optical-thin, 3-dimensional solar coronal loop centers from extreme ultraviolet and X-ray 2-dimensional images will be presented. These center splines are proxies of associated magnetic field lines. The project is pattern recognition problems in which there are no unique shapes or edges and in which photon and detector noise heavily influence the images. The study explores extraction techniques using: (1) linear feature recognition of local patterns (related to the inertia-tensor concept), (2) parametric space via the Hough transform, and (3) topological adaptive contours (snakes) that constrains curvature and continuity as possible candidates for digital loop detection schemes. We have developed synthesized images for the coronal loops to test the various loop identification algorithms. Since the topology of these solar features is dominated by the magnetic field structure, a first-order magnetic field approximation using multiple dipoles provides a priori information in the identification process. Results from both synthesized and solar images will be presented.

Lee, Jong K.↗
NASA NTRS1988

Infrared systems and components II; Proceedings of the Meeting, Los Angeles, CA, Jan. 14, 15, 1988

The present conference on the development status of IR electrooptics technologies discusses topics concerning such systems' materials and architectures, their modeling and data processing requirements, and advancements in IR automatic target recognition. Attention is given to active multispectral three-dimensional vision sensor image evaluation, a VHSIC processor-based real-time image tracker, the use of Al-alloy castings as diamond machining substrates for optical surfaces, a two-dimensional recognition range model, calculations of the optical blurring in IR systems, synthetic IR scene generation, and automatic target recognition evaluation.

Liaw, H. M.↗