La polarisation de la lumière est très souvent utilisée en imagerie pour caractériser certaines propriétés de la matière, ou pour mettre en évidence des zones qui ne seraient peu ou pas contrastées avec des caméras d'intensité classiques. Nous explorons le potentiel d'une nouvelle technique de polarimétrie, dite de « brisure d'orthogonalité », pour réaliser des acquisitions de manière simple, directe et à haute cadence. Cette technique d'imagerie par balayage laser repose sur l'emploi d'une source de lumière bi-fréquence bi-polarisation pour sonder les caractéristiques polarimétriques (notamment le dichroïsme ou anisotropie d'absorption) des échantillons imagés.Nous explorons la possibilité du déport de la mesure de « brisure d'orthogonalité » par fibre optique faiblement multimode pour le développement d'endoscopes polarimétriques. Un tel dispositif vise à fournir une méthode de diagnostic rapide pour analyser des tissus biologiques profonds tout en évitant le recours aux biopsies. Nous démontrons, théoriquement et expérimentalement, la compatibilité de cette approche avec un dispositif d'endoscopie commercial (fibre ou bundle multicœurs, légèrement multimodes) pourvu que le nombre de modes guidés soit inférieur à une dizaine. D'autre part, nous présentons la conception, la réalisation, la validation et l'exploitation d'un démonstrateur d'imagerie active par brisure d'orthogonalité dans le proche infrarouge. Ce dernier vise des applications défense de détection et/ou décamouflage de cibles. Après caractérisation des bruits dominants les signaux acquis, nous illustrons l'apport du démonstrateur pour la mise en évidence d'éléments dichroïques. Enfin, nous démontrons que la technique de brisure d'orthogonalité peut être avantageusement, et très simplement, adaptée pour mesurer sélectivement le dichroïsme, la biréfringence, et la dépolarisation, paramètres essentiels à la détection d'objets manufacturés (cibles). Ces trois modalités, lorsqu'elles sont conjuguées, offrent au démonstrateur des capacités d'identification. / Polarimetric imaging is a useful tool to characterize some matter properties, or to highlight regions slightly or not contrasted with intensity cameras. We investigate the capability of a novel polarimetric technique, namely the “orthogonality breaking technique”, to perform direct and straightforward measurements at high speed. Relying on the use of a dual-frequency dual-polarization light source, this imaging modality probes polarimetric features (dichroism, or absorption anisotropy) in imaged samples.We explore the potential to perform orthogonality breaking measurements through few mode optical fibers towards polarimetric endoscopy. Such an imaging device would greatly improve the diagnosis efficiency to analyze in-depth biologic tissues without biopsy surgery. We show, theoretically and experimentally, the compatibility of our approach with a commercial flexible endoscope (slightly multimode multicore fibers or bundle) provided that the number of guided modes remains inferior to a dozen.On the other hand, we describe the design, the development, the validation and the exploitation of an active near infrared imaging demonstrator based on the orthogonality breaking technique for defense target detection applications. After characterization of the acquired signals noise, we illustrate the imager capability to reveal dichroic elements. Finally, we demonstrate that the orthogonality breaking technique can be advantageously and straightforwardly tailored to address selectively the dichroism, the birefringence and the depolarization, which are core parameters for the detection of manufactured objects (targets). The combination of these three modalities grants an identification capability to the demonstrator.

This dissertation develops new approaches for detection and classification of buried radioactive materials. Different spectral transformation methods are proposed to effectively suppress noise and to better distinguish signal features in the transformed space. The contributions of this dissertation are detailed as follows. 1) Propose an unsupervised method for buried radioactive material detection. In the experiments, the original Reed-Xiaoli (RX) algorithm performs similarly as the gross count (GC) method; however, the constrained energy minimization (CEM) method performs better if using feature vectors selected from the RX output. Thus, an unsupervised method is developed by combining the RX and CEM methods, which can efficiently suppress the background noise when applied to the dimensionality-reduced data from principle component analysis (PCA). 2) Propose an approach for buried target detection and classification, which applies spectral transformation followed by noisejusted PCA (NAPCA). To meet the requirement of practical survey mapping, we focus on the circumstance when sensor dwell time is very short. The results show that spectral transformation can alleviate the effects from spectral noisy variation and background clutters, while NAPCA, a better choice than PCA, can extract key features for the following detection and classification. 3) Propose a particle swarm optimization (PSO)-based system to automatically determine the optimal partition for spectral transformation. Two PSOs are incorporated in the system with the outer one being responsible for selecting the optimal number of bins and the inner one for optimal bin-widths. The experimental results demonstrate that using variable bin-widths is better than a fixed bin-width, and PSO can provide better results than the traditional Powell’s method. 4) Develop parallel implementation schemes for the PSO-based spectral partition algorithm. Both cluster and graphics processing units (GPU) implementation are designed. The computational burden of serial version has been greatly reduced. The experimental results also show that GPU algorithm has similar speedup as cluster-based algorithm.

<p>Detection of dim point objects plays an important role in many imaging applications such as early warning systems, surveillance, astronomy, and microscopy. In satellite imaging, natural phenomena, such as clouds, can confound object detection methods. We propose an object detection method that uses spatial, spectral, and temporal information to reject detections that are not consistent with a moving object and achieve a high probability of detection with a low false alarm rate. We propose another method for dim object detection using convolutional neural networks (CNN). The method augments a conventional space-based detection processing chain with a lightweight CNN to improve detection performance. For evaluation of the performance of our proposed methods,</p> <p>we used a set of curated satellite images and generated receiver operating characteristics (ROC).</p> <p><br></p> <p>Most satellite images have adequate spatial resolution and signal-to-noise ratio (SNR) for the detection and localization of common large objects, such as buildings. In many applications, the spatial resolution of the imaging system is not enough to localize a point object or two closely-spaced objects (CSOs) that are described by only a few pixels (or less than one pixel). A low signal-to-noise ratio (SNR) increases the difficulty such as when the objects are dim. We describe a method to estimate the objects’ amplitudes and spatial locations with sub-pixel accuracy using non-linear optimization and information from multiple spectral bands. We also propose a machine</p> <p>learning method that minimizes a cost function derived from the maximum likelihood estimation of the observed image to determine an object’s sub-pixel spatial location and amplitude. We derive the Cramer-Rao Lower Bound and compare the proposed estimators’ variance with this bound.</p>

Polarimetric Synthetic Aperture Radar (PolSAR) has been used for various remote sensing applications since more information could be obtained in multiple polarizations. The overall objective of this thesis is to investigate urban area information extraction from PolSAR data with the following specific objectives: (1) to exploit polarimetric scattering model-based decomposition methods for urban areas, (2) to investigate effective methods for man-made target detection, (3) to develop edge detection and superpixel generation methods, and (4) to investigate urban area classification and segmentation. Paper 1 proposes a new scattering coherency matrix to model the cross-polarized scattering component from urban areas, which adaptively considers the polarization orientation angles of buildings. Thus, the HV scattering components from forests and oriented urban areas can be modelled respectively. Paper 2 presents two urban area decompositions using this scattering model. After the decomposition, urban scattering components can be effectively extracted. Paper 3 presents an improved man-made target detection method for PolSAR data based on nonstationarity and asymmetry. Reflection asymmetry was incorporate into the azimuth nonstationarity extraction method to improve the man-made target detection accuracy, i.e., removing the natural areas and detecting the small targets. In Paper 4, the edge detection of PolSAR data was investigated using SIRV model and Gauss-shaped filter. This detector can locate the edge pixels accurately with fewer omissions. This could be useful for speckle noise reduction, superpixel generation and others. Paper 5 investigates an unsupervised classification method for PolSAR data in urban areas. The ortho and oriented buildings can be discriminated very well. Paper 6 proposes an adaptive superpixel generation method for PolSAR images. The algorithm produces compact superpixels that can well adhere to image boundaries in both natural and urban areas. / Polarimetriska Synthetic Aperture Radar (PolSAR) har använts för olika fjärranalystillämpningar för, eftersom mer information kan erhållas från multipolarisad data. Det övergripande syftet med denna avhandling är att undersöka informationshämtning över urbana områden från PolSAR data med följande särskilda mål: (1) att utnyttja polarimetrisk spridningsmodellbaserade nedbrytningsmetoder för stadsområden, (2) att undersöka effektiva metoder för upptäckt av konstgjorda objekt, (3) att utveckla metoder som kantavkänning och superpixel generation, och (4) för att undersöka klassificering och segmentering av stadsområden. Artikel 1 föreslår en ny spridnings-koherens matris för att modellera korspolariserade spridningskomponent från tätorter, som adaptivt utvärderar polariseringsorienteringsvinkel av byggnader. Artikel 2 presenterar nedbrytningstekniken över två urbana områden med hjälp av denna spridningsmodell. Efter nedbrytningen kunde urbana spridningskomponenter effektivt extraheras. Artikel 3 presenterar en förbättrad detekteringsmetod för konstgjorda mål med PolSAR data baserade på icke-stationaritet och asymmetri. integrerades reflektionsasymmetri i icke-stationaritetsmetoden för att förbättra noggrannheten i upptäckten av konstgjorda föremål, dvs. att ta bort naturområden och upptäcka de små föremålen. I artikel 4 undersöktes kantdetektering av PolSAR data med hjälp av SIRV modell och ett Gauss-formad filter. Denna detektor kan hitta kantpixlarna noggrant med mindre utelämnande. Detta skulle den vara användbar för reduktion av brus, superpixel generation och andra. Artikel 5 utforskar en oövervakad klassificeringsmetod av PolSAR data över stadsområden. Orto- och orienterade byggnader kan särskiljas mycket väl. Baserat på artikel 4 föreslår artikel 6 en adaptiv superpixel generationensmetod för PolSAR data. Algoritmen producerar kompakta superpixels som kan kommer att följa bildgränser i både naturliga och stadsområden. / <p>QC 20160607</p>

Approved for public release, distribution is unlimited / Visual perception modeling is generally weak for game AI and computer generated forces (CGF), or agents, in computer games and military simulations. Several tricks and shortcuts are used in perceptual modeling. The results are, under certain conditions, unrealistic behaviors that negatively effect user immersion in games and call into question the validity of calculations in fine resolution military simulations. By determining what the computer-generated agent sees using methods similar to that used to generate the human players' screen view in 3- D virtual environments, we hope to present a method that can more accurately model human visual perception, specifically the major problem of a entity "hiding in plain sight" / Lieutenant, United States Navy

A fundamental challenge in target detection in hyperspectral imagery is spectral variability. In target detection applications, we are provided with a pure target signature; we do not have a collection of samples that characterize the spectral variability of the target. Another problem is that the performance of stochastic detection algorithms such as the spectral matched filter can be detrimentally affected by the assumptions of multivariate normality of the data, which are often violated in practical situations. We address the challenge of lack of training samples by creating two models to characterize the target class spectral variability --the first model makes no assumptions regarding inter-band correlation, while the second model uses a first-order Markovbased scheme to exploit correlation between bands. Using these models, we present two techniques for meeting these challenges-the kernel-based support vector data description (SVDD) and spectral fringe-adjusted joint transform correlation (SFJTC). We have developed an algorithm that uses the kernel-based SVDD for use in full-pixel target detection scenarios. We have addressed optimization of the SVDD kernel-width parameter using the golden-section search algorithm for unconstrained optimization. We investigated a proper number of signatures N to generate for the SVDD target class and found that only a small number of training samples is required relative to the dimensionality (number of bands). We have extended decision-level fusion techniques using the majority vote rule for the purpose of alleviating the problem of selecting a proper value of s 2 for either of our target variability models. We have shown that heavy spectral variability may cause SFJTC-based detection to suffer and have addressed this by developing an algorithm that selects an optimal combination of the discrete wavelet transform (DWT) coefficients of the signatures for use as features for detection. For most scenarios, our results show that our SVDD-based detection scheme provides low false positive rates while maintaining higher true positive rates than popular stochastic detection algorithms. Our results also show that our SFJTC-based detection scheme using the DWT coefficients can yield significant detection improvement compared to use of SFJTC using the original signatures and traditional stochastic and deterministic algorithms.

In this study, novel methods for high-resolution radar target detection in non-Gaussian clutter environment are proposed. In solution of the problem, two approaches are used: Non-coherent detection that operates on the envelope-detected signal for thresholding and coherent detection that performs clutter suppression, Doppler processing and thresholding at the same time. The proposed non-coherent detectors, which are designed to operate in non-Gaussian and range-heterogeneous clutter, yield higher performance than the conventional methods that were designed either for Gaussian clutter or heterogeneous clutter. The proposed coherent detector exploits the information in all the range cells and pulses and performs the clutter reduction and thresholding simultaneously. The design is performed for uncorrelated, partially correlated and fully correlated clutter among range cells. The performance analysis indicates the superiority of the designed methods over the classical ones, in fully correlated and partially correlated situations. In addition, by design of detectors for multiple targets and making corrections to the conventional methods, the target-masking problem of the classical detectors is alleviated.

In this thesis study, an overview of MIMO radar is presented. The differences in radar cross section, channel and received signal models in different MIMO radar configurations are examined. The performance improvements that can be achieved by the use of waveform diversity in coherent MIMO radar and by the use of angular diversity in statistical MIMO radar are investigated. The optimal detector under Neyman-Pearson criterion for Coherent MIMO radar when the interfering signal is white Gaussian noise is developed. Detection performance of phased array radar, coherent MIMO radar and Statistical MIMO radar are compared through numerical simulations. A detector for MIMO radar that contains the space time codes explicitly is also examined.

The goal of this thesis study is to investigate the Ambiguity Function Technique for target detection in phase-coded continuous wave radar. Also, phase shift keying techniques are examined in detail. Continuous Wave (CW) Radars, which are also known as Low Probability of Intercept (LPI) radars, emit continuous signals in time which are modulated by either frequency modulation or phase modulation techniques. Modulation of the transmitted radar signal is needed to estimate both the range and the radial velocity of the detected targets. In this thesis, Phase Shift Keying (PSK) techniques such as the Barker codes, Frank codes, P1, P2, P3, P4 codes will be employed for radar signal modulation. The use of Ambiguity Function, which is a non-linear Time- Frequency Representation (TFR), for target detection will be investigated in phasecoded CW radars for different target scenarios.

The move from Standard Definition (SD) to High Definition (HD) represents a six times increases in data, which needs to be processed. With expanding resolutions and evolving compression, there is a need for high performance with flexible architectures to allow for quick upgrade ability. The technology advances in image display resolutions, advanced compression techniques, and video intelligence. Software implementation of these systems can attain accuracy with tradeoffs among processing performance (to achieve specified frame rates, working on large image data sets), power and cost constraints. There is a need for new architectures to be in pace with the fast innovations in video and imaging. It contains dedicated hardware implementation of the pixel and frame rate processes on Field Programmable Gate Array (FPGA) to achieve the real-time performance. The following outlines the contributions of the dissertation. (1) We develop a target detection system by applying a novel running average mean threshold (RAMT) approach to globalize the threshold required for background subtraction. This approach adapts the threshold automatically to different environments (indoor and outdoor) and different targets (humans and vehicles). For low power consumption and better performance, we design the complete system on FPGA. (2) We introduce a safe distance factor and develop an algorithm for occlusion occurrence detection during target tracking. A novel mean-threshold is calculated by motion-position analysis. (3) A new strategy for gesture recognition is developed using Combinational Neural Networks (CNN) based on a tree structure. Analysis of the method is done on American Sign Language (ASL) gestures. We introduce novel point of interests approach to reduce the feature vector size and gradient threshold approach for accurate classification. (4) We design a gesture recognition system using a hardware/ software co-simulation neural network for high speed and low memory storage requirements provided by the FPGA. We develop an innovative maximum distant algorithm which uses only 0.39% of the image as the feature vector to train and test the system design. Database set gestures involved in different applications may vary. Therefore, it is highly essential to keep the feature vector as low as possible while maintaining the same accuracy and performance