Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Feature Extraction”

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 325 records · Page 18
All sourcesNASA NTRSDOE OSTI
NASA NTRS1976

Classification improvement by optimal dimensionality reduction when training sets are of small size

A computer simulation was performed to test the conjecture that, when the sizes of the training sets are small, classification in a subspace of the original data space may give rise to a smaller probability of error than the classification in the data space itself; this is because the gain in the accuracy of estimation of the likelihood functions used in classification in the lower dimensional space (subspace) offsets the loss of information associated with dimensionality reduction (feature extraction). A number of pseudo-random training and data vectors were generated from two four-dimensional Gaussian classes. A special algorithm was used to create an optimal one-dimensional feature space on which to project the data. When the sizes of the training sets are small, classification of the data in the optimal one-dimensional space is found to yield lower error rates than the one in the original four-dimensional space.

Starks, S. A.↗
NASA NTRS1990

Real-time image processing II; Proceedings of the Meeting, Orlando, FL, Apr. 16-18, 1990

The present conference discusses topics in the fields of feature extraction and implementation, filter and correlation algorithms, optical correlators, high-level algorithms, and digital image processing for ranging and remote driving. Attention is given to a nonlinear filter derived from topological image features, IR image segmentation through iterative thresholding, orthogonal subspaces for correlation masking, composite filter trees and image recognition via binary search, and features of matrix-coherent optical image processing. Also discussed are multitarget tracking via hybrid joint transform correlator, binary joint Fourier transform correlator considerations, global image processing operations on parallel architectures, real-time implementation of a differential range finder, and real-time binocular stereo range and motion detection.

Juday, Richard D.↗
DOE OSTI2019

Determining moisture content of Traditional Chinese Medicines using a near-infrared LED-based moisture content sensor with spectrum analysis

In Traditional Chinese Medicine (TCM), moisture content (MC) inspection using voltage level or a lock-in amplifier, with a features-based near-infrared light-emitting diode sensor is possible. Abundant absorption response in a wide frequency range and information in both the time-domain and frequency-domain can be obtained, with the square excitation method. However, planeness, heterogeneous particle sizes, and measurement noise negatively affect the time-domain feature during MC inspection. In this work, spectrum analysis (SA) for feature extraction is investigated, to precisely measure the MC of Dutong pills. The principal components for low and high frequencies, are aimed at counteracting the dimensional methods, which are used for supplying the lower dimensional features. The SA experimental results demonstrate that the low frequency components, ranging from the base to the 3rd harmonic components, have a high linear correlation with the MC. The calibration and validation experimental results reveal that the MC can be a monotonic function about the SA amplitudes or first principal component (PC1), with the proposed sensor. This method has the potential to be used for gauging automatic, in situ MCs, particularly for quality control and Process Analytical Technology in TCM. Future research opportunities include improving accuracy and application of the method under field conditions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2025

Laser powder bed fusion parameter estimation with k-NN

Abstract Laser powder bed fusion (L-PBF) is a technique within additive manufacturing that uses a high power density laser to build parts from fused powdered metal alloy. This technology is well equipped to produce complex parts with otherwise impossible features, such as hidden voids or lattice structures. Alongside capability, reliability and quality are key characteristics considered when choosing a manufacturing method, and these are gaining attention as this method becomes more prevalent in industry. One main indicator of a stable L-PBF process is consistent melt pool geometry, and the properties of which are likely to determine the quality of the part produced. As computing power and sensing technologies become more advanced, this melt pool geometry could be studied in real time. This work addresses the challenge by leveraging a k-nearest neighbor (k-NN) model to identify key features within melt pool imagery and predict the energy density. The k-NN model was trained on data provided by the National Institute of Standards and Technology (NIST). Data preprocessing was performed on the images to extract features that were used in the k-NN model. This approach was used to accurately infer the energy density of unseen layers within the same part. The algorithm was subsequently tested with unique scan strategies and found to reasonably estimate the energy density of different parts. A fivefold cross validation found the algorithm to be consistently predicting the class of 91.4% of the in situ melt pool images.

Jung, Patrick (ORCID:0000000267890859)↗
NASA NTRS1994

Unsupervised texture image segmentation by improved neural network ART2

We here propose a segmentation algorithm of texture image for a computer vision system on a space robot. An improved adaptive resonance theory (ART2) for analog input patterns is adapted to classify the image based on a set of texture image features extracted by a fast spatial gray level dependence method (SGLDM). The nonlinear thresholding functions in input layer of the neural network have been constructed by two parts: firstly, to reduce the effects of image noises on the features, a set of sigmoid functions is chosen depending on the types of the feature; secondly, to enhance the contrast of the features, we adopt fuzzy mapping functions. The cluster number in output layer can be increased by an autogrowing mechanism constantly when a new pattern happens. Experimental results and original or segmented pictures are shown, including the comparison between this approach and K-means algorithm. The system written in C language is performed on a SUN-4/330 sparc-station with an image board IT-150 and a CCD camera.

Wang, Zhiling↗
NASA NTRS1992

Feature detection in satellite images using neural network technology

A feasibility study of automated classification of satellite images is described. Satellite images were characterized by the textures they contain. In particular, the detection of cloud textures was investigated. The method of second-order gray level statistics, using co-occurrence matrices, was applied to extract feature vectors from image segments. Neural network technology was employed to classify these feature vectors. The cascade-correlation architecture was successfully used as a classifier. The use of a Kohonen network was also investigated but this architecture could not reliably classify the feature vectors due to the complicated structure of the classification problem. The best results were obtained when data from different spectral bands were fused.

Augusteijn, Marijke F.↗
DOE OSTI2015

The value of nodal information in predicting lung cancer relapse using 4DPET/4DCT

Purpose: There is evidence that computed tomography (CT) and positron emission tomography (PET) imaging metrics are prognostic and predictive in nonsmall cell lung cancer (NSCLC) treatment outcomes. However, few studies have explored the use of standardized uptake value (SUV)-based image features of nodal regions as predictive features. The authors investigated and compared the use of tumor and node image features extracted from the radiotherapy target volumes to predict relapse in a cohort of NSCLC patients undergoing chemoradiation treatment. Methods: A prospective cohort of 25 patients with locally advanced NSCLC underwent 4DPET/4DCT imaging for radiation planning. Thirty-seven image features were derived from the CT-defined volumes and SUVs of the PET image from both the tumor and nodal target regions. The machine learning methods of logistic regression and repeated stratified five-fold cross-validation (CV) were used to predict local and overall relapses in 2 yr. The authors used well-known feature selection methods (Spearman’s rank correlation, recursive feature elimination) within each fold of CV. Classifiers were ranked on their Matthew’s correlation coefficient (MCC) after CV. Area under the curve, sensitivity, and specificity values are also presented. Results: For predicting local relapse, the best classifier found had a mean MCC of 0.07 and was composed of eight tumor features. For predicting overall relapse, the best classifier found had a mean MCC of 0.29 and was composed of a single feature: the volume greater than 0.5 times the maximum SUV (N). Conclusions: The best classifier for predicting local relapse had only tumor features. In contrast, the best classifier for predicting overall relapse included a node feature. Overall, the methods showed that nodes add value in predicting overall relapse but not local relapse.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2021

An adaptive adversarial domain adaptation approach for corn yield prediction

Recently, statistical machine learning and deep learning methods have been widely explored for corn yield prediction. Though successful, machine learning models generated within a specific spatial domain often lose their validity when directly applied to new regions. To address this issue, we designed an unsupervised adaptive domain adversarial neural network (ADANN). Specifically, through domain adversarial training, the ADANN model reduced the impact of domain shift by projecting data from different domains into the same subspace. Also, the ADANN model was designed to be trained in an adaptive way, which guaranteed the model can learn the domain-invariant features and perform accurate yield prediction simultaneously. Informative variables including time-series vegetation indices and sequential weather observations were first collected from multiple data sources and aggregated to the county level. Then, we trained the ADANN model with the extracted features and corresponding reported county-level corn yield from the U.S. Department of Agriculture (USDA). Finally, the trained model was evaluated in four testing years 2016–2019. The U.S. corn belt was used as the study area and counties under study were grouped into two diverse ecological regions. Overall, the experimental results showed that the developed ADANN model had better performance than three other state-of-the-art machine learning models in both local experiments (train and test in the same region) and transfer experiments (train and test in different regions). As the first study using adversarial learning for crop yield prediction, this research demonstrates a novel solution for improving model transferability on crop yield prediction.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2017

Improved neutron-gamma discrimination for a 3 He neutron detector using subspace learning methods

3 He gas based neutron linear-position-sensitive detectors (LPSDs) have been applied for many neutron scattering instruments. Traditional Pulse-Height Analysis (PHA) for Neutron-Gamma Discrimination (NGD) resulted in the neutron-gamma efficiency ratio on the orders of 10 5 -10 6 . The NGD ratios of 3 He detectors need to be improved for even better scientific results from neutron scattering. Digital Signal Processing (DSP) analyses of waveforms were proposed for obtaining better NGD ratios, based on features extracted from rise-time, pulse amplitude, charge integration, a simplified Wiener filter, and the cross-correlation between individual and template waveforms of neutron and gamma events. Fisher linear discriminant analysis (FLDA) and three multivariate analyses (MVAs) of the features were performed. The NGD ratios are improved by about 10 2 -10 3 times compared with the traditional PHA method. Finally, our results indicate the NGD capabilities of 3 He tube detectors can be significantly improved with subspace-learning based methods, which may result in a reduced data-collection time and better data quality for further data reduction.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2019

Exploring End-to-end Deep Learning Applications for Event Classification at CMS

An essential part of new physics searches at the Large Hadron Collider (LHC) at CERN involves event classification, or distinguishing potential signal events from those coming from background processes. Current machine learning techniques accomplish this using traditional hand-engineered features like particle 4-momenta, motivated by our understanding of particle decay phenomenology. While such techniques have proven useful for simple decays, they are highly dependent on our ability to model all aspects of the phenomenology and detector response. Meanwhile, powerful deep learning algorithms are capable of not only training on high-level features, but of performing feature extraction. In computer vision, convolutional neural networks have become the state-of-the-art for many applications. Motivated by their success, we apply deep learning algorithms to low-level detector data from the 2012 CMS Simulated Open Data to directly learn useful features, in what we call, end-to-end event classification. We demonstrate the power of this approach in the context of a physics search and offer solutions to some of the inherent challenges, such as image construction, image sparsity, combining multiple sub-detectors, and de-correlating the classifier from the search observable, among others.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2019

Vehicle detection based on visual attention mechanism and adaboost cascade classifier in intelligent transportation systems

Robust and efficient vehicle detection is an essential task in intelligent transportation systems (ITS). Unfortunately, due to a great diversity of vehicle profiles and outdoor illumination conditions, it is a challenge to detect vehicles effectively. This paper proposes a method for high-performance vehicle detection based on visual attention mechanism and AdaBoost cascade classifier. Our method constructs the structural Haar features and extracts the features of samples using structural Haar features and trains an AdaBoost cascade classifier. Then we use the visual attention mechanism to extract the target candidate region. At last, we generate detecting sub-windows in the candidate region and discriminate them with the cascade classifier to realize vehicle detection. We compare the performance of this method against two variants, one using MB–LBP features and another using Haar features. The experimental results demonstrate satisfactory performance for the proposed method in term of training speed, detecting speed and detecting accuracy.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
NASA NTRS2001

Road Extraction from AVIRIS Using Spectral Mixture and Q-Tree Filter Techniques

Accurate road location and condition information are of primary importance in road infrastructure management. Additionally, spatially accurate and up-to-date road networks are essential in ambulance and rescue dispatch in emergency situations. However, accurate road infrastructure databases do not exist for vast areas, particularly in areas with rapid expansion. Currently, the US Department of Transportation (USDOT) extends great effort in field Global Positioning System (GPS) mapping and condition assessment to meet these informational needs. This methodology, though effective, is both time-consuming and costly, because every road within a DOT's jurisdiction must be field-visited to obtain accurate information. Therefore, the USDOT is interested in identifying new technologies that could help meet road infrastructure informational needs more effectively. Remote sensing provides one means by which large areas may be mapped with a high standard of accuracy and is a technology with great potential in infrastructure mapping. The goal of our research is to develop accurate road extraction techniques using high spatial resolution, fine spectral resolution imagery. Additionally, our research will explore the use of hyperspectral data in assessing road quality. Finally, this research aims to define the spatial and spectral requirements for remote sensing data to be used successfully for road feature extraction and road quality mapping. Our findings will facilitate the USDOT in assessing remote sensing as a new resource in infrastructure studies.

Gardner, Margaret E.↗
DOE OSTI2016

SU-D-207B-02: Early Grade Classification in Meningioma Patients Combining Radiomics and Semantics Data

Purpose: The clinical management of meningioma is guided by its grade and biologic behavior. Currently, diagnosis of tumor grade follows surgical resection and histopathologic review. Reliable techniques for pre-operative determination of tumor behavior are needed. We investigated the association between imaging features extracted from preoperative gadolinium-enhanced T1-weighted MRI and meningioma grade. Methods: We retrospectively examined the pre-operative MRI for 139 patients with de novo WHO grade I (63%) and grade II (37%) meningiomas. We investigated the predictive power of ten semantic radiologic features as determined by a neuroradiologist, fifteen radiomic features, and tumor location. Conventional (volume and diameter) imaging features were added for comparison. AUC was computed for continuous and χ{sup 2} for discrete variables. Classification was done using random forest. Performance was evaluated using cross validation (1000 iterations, 75% training and 25% validation). All p-values were adjusted for multiple testing. Results: Significant association was observed between meningioma grade and tumor location (p<0.001) and two semantic features including intra-tumoral heterogeneity (p<0.001) and overt hemorrhage (p=0.01). Conventional (AUC 0.61–0.67) and eleven radiomic (AUC 0.60–0.70) features were significant from random (p<0.05, Noether test). Median AUC values for classification of tumor grade were 0.57, 0.71, 0.72 and 0.77 respectively for conventional, radiomic, location, and semantic features after using random forest. By combining all imaging data (semantic, radiomic, and location), the median AUC was 0.81, which offers superior predicting power to that of conventional imaging descriptors for meningioma as well as radiomic features alone (p<0.05, permutation test). Conclusion: We demonstrate a strong association between radiologic features and meningioma grade. Pre-operative prediction of tumor behavior based on imaging features offers promise for guiding personalized medicine and improving patient management.

60 APPLIED LIFE SCIENCES↗
NASA NTRS1991

Optical implementation of neocognitron and its applications to radar signature discrimination

A feature-extraction-based optoelectronic neural network is introduced. The system implementation approach applies the principle of the neocognitron paradigm first introduced by Fukushima et al. (1983). A multichannel correlator is used as a building block of a generic single layer of the neocognitron for shift-invariant feature correlation. Multilayer processing is achieved by iteratively feeding back the output of the feature correlator to the input spatial light modulator. Successful pattern recognition with intraclass fault tolerance and interclass discrimination is achieved using this optoelectronic neocognitron. Detailed system analysis is described. Experimental demonstration of radar signature processing is also provided.

Chao, Tien-Hsin↗
NASA NTRS2009

Processing Electromyographic Signals to Recognize Words

A recently invented speech-recognition method applies to words that are articulated by means of the tongue and throat muscles but are otherwise not voiced or, at most, are spoken sotto voce. This method could satisfy a need for speech recognition under circumstances in which normal audible speech is difficult, poses a hazard, is disturbing to listeners, or compromises privacy. The method could also be used to augment traditional speech recognition by providing an additional source of information about articulator activity. The method can be characterized as intermediate between (1) conventional speech recognition through processing of voice sounds and (2) a method, not yet developed, of processing electroencephalographic signals to extract unspoken words directly from thoughts. This method involves computational processing of digitized electromyographic (EMG) signals from muscle innervation acquired by surface electrodes under a subject's chin near the tongue and on the side of the subject s throat near the larynx. After preprocessing, digitization, and feature extraction, EMG signals are processed by a neural-network pattern classifier, implemented in software, that performs the bulk of the recognition task as described.

Jorgensen, C. C.↗
DOE OSTI2018

Unobtrusive Sensor-Based Occupancy Facing Direction Detection and Tracking Using Advanced Machine Learning Algorithms

Facing direction detection plays a critical role in human-computer interaction and has a wide application in surveillance systems, driving awareness recognition, smart home appliances, computer games, and so on. Current detection methods are mainly focused on extracting specific patterns from user's optical images, which raises concerns on privacy invasion and these detection techniques do not usually work in a dark environment. To address these concerns, here this paper proposes an activity recognition system guided by an unobtrusive sensor (ARGUS). By using a low pixel infrared thermopile array sensor, ARGUS is capable of identifying five facing directions (left 45°/90°, right 45°/90°, and front) through the support vector machine classifier. Also two feature extraction methods are compared. One is manually-defined and the other is based on a pre-trained convolutional neural network (CNN) model. The facing direction detection accuracy resulting from manually-defined features reaches 85.3%, 90.6%, and 85.2% at detection distance of 0.6, 1.2, and 1.8 m, respectively. The level of accuracy resulted from using pre-trained CNN features demonstrates a much more reliable performance (89.1%, 95.3%, and 95.1% at distance of 0.6, 1.2, and 1.8 m, respectively). In addition, ARGUS has been successfully applied for occupancy tracking with a root mean square error of 0.19 m.

42 ENGINEERING↗
NASA NTRS1979

Low-level processing for real-time image analysis

A system that detects object outlines in television images in real time is described. A high-speed pipeline processor transforms the raw image into an edge map and a microprocessor, which is integrated into the system, clusters the edges, and represents them as chain codes. Image statistics, useful for higher level tasks such as pattern recognition, are computed by the microprocessor. Peak intensity and peak gradient values are extracted within a programmable window and are used for iris and focus control. The algorithms implemented in hardware and the pipeline processor architecture are described. The strategy for partitioning functions in the pipeline was chosen to make the implementation modular. The microprocessor interface allows flexible and adaptive control of the feature extraction process. The software algorithms for clustering edge segments, creating chain codes, and computing image statistics are also discussed. A strategy for real time image analysis that uses this system is given.

Eskenazi, R.↗