Open Access. Powered by Scholars. Published by Universities.®

Engineering Commons

Open Access. Powered by Scholars. Published by Universities.®

Electrical and Computer Engineering

Air Force Institute of Technology

Pattern recognition systems

Articles 1 - 15 of 15

Full-Text Articles in Engineering

Towards The Mitigation Of Correlation Effects In The Analysis Of Hyperspectral Imagery With Extension To Robust Parameter Design, Jason P. Williams Sep 2012

Towards The Mitigation Of Correlation Effects In The Analysis Of Hyperspectral Imagery With Extension To Robust Parameter Design, Jason P. Williams

Theses and Dissertations

Standard anomaly detectors and classifiers assume data to be uncorrelated and homogeneous, which is not inherent in Hyperspectral Imagery (HSI). To address the detection difficulty, a new method termed Iterative Linear RX (ILRX) uses a line of pixels which shows an advantage over RX, in that it mitigates some of the effects of correlation due to spatial proximity; while the iterative adaptation from Iterative Linear RX (IRX) simultaneously eliminates outliers. In this research, the application of classification algorithms using anomaly detectors to remove potential anomalies from mean vector and covariance matrix estimates and addressing non-homogeneity through cluster analysis, both of …


Computer Aided Multi-Data Fusion Dismount Modeling, Juan L. Morales Mar 2012

Computer Aided Multi-Data Fusion Dismount Modeling, Juan L. Morales

Theses and Dissertations

Recent research efforts strive to address the growing need for dismount surveillance, dismount tracking and characterization. Current work in this area utilizes hyperspectral and multispectral imaging systems to exploit spectral properties in order to detect areas of exposed skin and clothing characteristics. Because of the large bandwidth and high resolution, hyperspectral imaging systems pose great ability to characterize and detect dismounts. A multi-data dismount modeling system where the development and manipulation of dismount models is a necessity. This thesis demonstrates a computer aided multi-data fused dismount model, which facilitates studies of dismount detection, characterization and identification. The system is created …


Applying Image Matching To Video Analysis, Adam J. Behring Sep 2010

Applying Image Matching To Video Analysis, Adam J. Behring

Theses and Dissertations

Dealing with the volume of multimedia collected on a daily basis for intelligence gathering and digital forensics investigations requires significant manual analysis. A component of this problem is that a video may be reanalyzed that has already been analyzed. Identifying duplicate video sequences is difficult due to differences in videos of varying quality and size. This research uses a kd-tree structure to increase image matching speed. Keypoints are generated and added to a kd-tree of a large dimensionality (128 dimensions). All of the keypoints for the set of images are used to construct a global kd-tree, which allows nearest neighbor …


Flexible Computing Architecture For Real Time Skin Detection, Matthew P. Hornung Mar 2010

Flexible Computing Architecture For Real Time Skin Detection, Matthew P. Hornung

Theses and Dissertations

In both the Air Force and Search and Rescue Communities, there is a current need to detect and characterize persons. Existing methods use red-green-blue (RGB) imagery, but produce high false alarm rates. New technology in multi-spectral skin detection is better than the existing RGB methods, but lacks a control and processing architecture to make them efficient for real time problems. We hypothesize that taking a minimalistic approach to the software design, we can perform image preprocessing, feature computation, and skin detection in real time. A number of applications require accurate detection and characterization of persons, human measurement and signature intelligence …


Improved Multispectral Skin Detection And Its Application To Search Space Reduction For Dismount Detection Based On Histograms Of Oriented Gradients, Adam L. Brooks Mar 2010

Improved Multispectral Skin Detection And Its Application To Search Space Reduction For Dismount Detection Based On Histograms Of Oriented Gradients, Adam L. Brooks

Theses and Dissertations

Due to the general shift from conventional warfare to terrorism and urban warfare by enemies of the United States in the late 20th Century, locating and tracking individuals of interest have become critically important. Dismount detection and tracking are vital to provide security and intelligence in both combat and homeland defense scenarios including base defense, combat search and rescue (CSAR), and border patrol. This thesis focuses on exploiting recent advances in skin detection research to reliably detect dismounts in a scene. To this end, a signal-plus-noise model is developed to map modeled skin spectra to the imaging response of an …


Multi-Dimensional Classification Algorithm For Automatic Modulation Recognition, Ouail Albairat Mar 2007

Multi-Dimensional Classification Algorithm For Automatic Modulation Recognition, Ouail Albairat

Theses and Dissertations

This thesis proposes an approach for modulation classification using existing features in a more efficient way. The Multi-Dimensional Classification Algorithm (MDCA) treats features extracted from signals of interest as elements with irrelevant identities, hence eliminating any dependence of the classifier on any particular feature. This design enables the use of any number of features, and the MDCA algorithm provides the capability to classify modulations in higher dimensions. The use of multiple features requires an equal number of data dimensions, and thus classification in as high a dimensional space as possible can improve final classification results. Finally, the MDCA algorithm uses …


Automatic Target Recognition User Interface Tool, David A. Kerns Mar 2007

Automatic Target Recognition User Interface Tool, David A. Kerns

Theses and Dissertations

A computer tool to aid in selecting the best Automatic Target Recognition (ATR) algorithm is developed. The program considers many quantifiers, accepts user-defined parameters, allows for changes in the operational environment and presents results in a meaningful way. It is written for Microsoft Excel. An ATR algorithm assigns a class label to a recognized target. General designations can include "Friend" and "Foe." The error of designating "Friend" as "Foe" as well as "Foe" as "Friend" comes with a high cost. Studying each algorithm's error can minimize this cost. Receiver Operating Characteristic (ROC) curves provide only information on the probabilities given …


Daytime Detection Of Space Objects, Alistair D. Funge Mar 2005

Daytime Detection Of Space Objects, Alistair D. Funge

Theses and Dissertations

Space Situational Awareness (SSA) requires repeated object updates for orbit accuracy. Detection of unknown objects is critical. A daytime model was developed that evaluated sun flares and assessed thermal emissions from space objects. Iridium satellites generate predictable sun glints. These were used as a model baseline for daytime detections. Flares and space object thermal emissions were examined for daytime detection. A variety of geometric, material and atmospheric characteristics affected this daytime detection capability. In a photon noise limited mode, simulated Iridium flares were detected. The peak Signal-to- Noise Ratios (SNR) were 6.05e18, 9.63e5, and 1.65e7 for the nighttime, daytime and …


Modification Of A Modulation Recognition Algorithm To Enable Multi-Carrier Recognition, Angela M. Waters Mar 2005

Modification Of A Modulation Recognition Algorithm To Enable Multi-Carrier Recognition, Angela M. Waters

Theses and Dissertations

Modulation recognition is important for both military and commercial communication applications, particularly in cases where enhanced situation awareness and/or channel assessment is required to mitigate intentional or collateral interference. Modulation recognition via template matching or statistical analysis is a key aspect of non-cooperative (nonmatched filtering) signal interception, classification, and exploitation. This research concerns the evaluation and modification of a conventional Digitally Modulated Signal Recognition Algorithm (DMRA) to enable multi-carrier, Orthogonal Frequency Division Multiplexing (OFDM), waveform recognition. The original DMRA architecture was developed to classify communication signals for three fundamental data modulations, i.e., Amplitude Shift Keying (ASK), Phase Shift Keying (PSK), …


Target Recognition Using Late-Time Returns From Ultra-Wideband, Short-Pulse Radar, Kenneth J. Pascoe Jun 2004

Target Recognition Using Late-Time Returns From Ultra-Wideband, Short-Pulse Radar, Kenneth J. Pascoe

Theses and Dissertations

The goal of this research is to develop algorithms that recognize targets by exploiting properties in the late-time resonance induced by ultra-wide band radar signals. A new variant of the Matrix Pencil Method algorithm is developed that identifies complex resonant frequencies present in the scattered signal. Kalman filters are developed to represent the dynamics of the signals scattered from several target types. The Multiple Model Adaptive Estimation algorithm uses the Kalman filters to recognize targets. The target recognition algorithm is shown to be successful in the presence of noise. The performance of the new algorithms is compared to that of …


Target Recognition Using Linear Classification Of High Range Resolution Radar Profiles, Ricardo A. Diaz Mar 2004

Target Recognition Using Linear Classification Of High Range Resolution Radar Profiles, Ricardo A. Diaz

Theses and Dissertations

High Range Resolution (HRR) radar profiles map three-dimensional target characteristics onto one-dimensional signals that represent reflected radar intensity along target extent. In this thesis, second through fourth statistical moments are extracted from HRR profiles and input to Fisher Linear Discriminant (FLD) classifiers. An iterative classification process is applied that gradually minimizes required a priori knowledge about the target data. It is found that the second through fourth statistical moments of HRR profiles are useful features in the FLD classification of dissimilar targets and they provide reasonable discrimination of similar targets. Greater than 69% correct classification for two-target scenarios and greater …


Classification Of Radar Targets Using Invariant Features, Gregory J. Meyer Apr 2003

Classification Of Radar Targets Using Invariant Features, Gregory J. Meyer

Theses and Dissertations

Automatic target recognition ATR using radar commonly relies on modeling a target as a collection of point scattering centers, Features extracted from these scattering centers for input to a target classifier may be constructed that are invariant to translation and rotation, i.e., they are independent of the position and aspect angle of the target in the radar scene. Here an iterative approach for building effective scattering center models is developed, and the shape space of these models is investigated. Experimental results are obtained for three-dimensional scattering centers compressed to nineteen-dimensional feature sets, each consisting of the singular values of the …


Space Object Identification Using Feature Space Trajectory Neural Networks, Neal W. Bruegger Mar 1997

Space Object Identification Using Feature Space Trajectory Neural Networks, Neal W. Bruegger

Theses and Dissertations

The Feature Space Trajectory Neural Network (FSTNN) is a simple yet powerful pattern recognition tool developed by Neiberg and Casasent for use in an Automatic Target Recognition System. Since the FSTNN was developed, it has been used on various problems including speaker identification and space object identification. However, in these types of problems, the test set represents time series data rather than an independent set of points. Since the distance metric of the standard FSTNN treats each test point independently without regard to its position in the sequence, the FSTNN can yield less than optimal results in these problems. Two …


Space Object Identification Using Spatio-Temporal Pattern Recognition, Gary W. Brandstrom Dec 1995

Space Object Identification Using Spatio-Temporal Pattern Recognition, Gary W. Brandstrom

Theses and Dissertations

This thesis is part of a research effort to automate the task of characterizing space objects or satellites based on a sequence of images. The goal is to detect space object anomalies. Two algorithms are considered - the feature space trajectory neural network (FST NN) and hidden Markov model (HMM) classifier. The FST NN was first presented by Leonard Neiberg and David P. Casasent in 1994 as a target identification tool. Kenneth H. Fielding and Dennis W. Ruck recently applied the hidden Markov model classifier to a 3D moving light display identification problem and a target recognition problem, using time …


Spatio-Temporal Pattern Recognition Using Hidden Markov Models, Kenneth H. Fielding Jun 1994

Spatio-Temporal Pattern Recognition Using Hidden Markov Models, Kenneth H. Fielding

Theses and Dissertations

A new spatio-temporal method for identifying 3D objects found in 2D image sequences is presented. The Hidden Markov Model technique is used as a spatio-temporal classification algorithm to identify 3D objects by the temporal changes in observed shape features. A new information theoretic argument is developed that proves identifying objects based on image sequences can lead to higher classification accuracies than single look methods. A new distance measure is proposed that analyzes the performance of Hidden Markov Models in a multi-class pattern recognition problem. A three class problem identifying moving light display objects provides experimental verification of the sequence processing …