Notre planète est pour l’essentiel recouverte par les mers et les océans, or notre connaissance des fonds marins est très inférieure à celle que nous possédons sur les terres émergées. Dans ce mémoire, nous cherchons à concevoir un système dédié à la cartographie thématique à grande échelle pour obtenir à la demande un nuage de points dense représentatif d’une scène sous-marine ou subaquatique par reconstruction tridimensionnelle. Le caractère complexe de ce type de système nous amène à privilégier une approche délibérément transversale. Nous nous intéresserons en particulier aux problématiques posées par l’étude à l’échelle des individus de petites zones en eau peu profonde. Les premières concernent l’acquisition in situ efficace de couples stéréoscopiques avec une logistique adaptée à la taille des zones observées : nous proposons pour cela un microsystème agile, peu coûteux et suffisamment automatisé pour fournir des données reproductibles et comparables. Les secondes portent sur l’extraction fiable de l’information tridimensionnelle à partir des données acquises : nous exposons les algorithmes que nous avons élaborés pour prendre en compte les caractéristiques particulières du milieu aquatique (dynamisme, propagation difficile des ondes électromagnétiques, etc.). Nous abordons donc en détail dans ce mémoire les problèmes d’appariement dense, d’étalonnage, d’acquisition in situ, de recalage et de redondance des données rencontrés dans le milieu sous-marin. / Our planet is mostly covered by seas and oceans. However, our knowledge of the seabed is far more restricted than that of land surface. In this thesis, we seek to develop a system dedicated to precise thematic mapping to obtain a dense point cloud of an underwater area on demand by using three-dimensional reconstruction. The complex nature of this type of system leads us to favor a multidisciplinary approach. We will examine in particular the issues raised by studying small shallow water areas on the scale of individual objects. The first problems concern the effective in situ acquisition of stereo pairs with logistics adapted to the sizes of the observed areas: for this, we propose an agile, affordable microsystem which is sufficiently automated to provide reproducible and comparable data. The second set of problems relates to the reliable extraction of three-dimensional information from the acquired data: we outline the algorithms we have developed to take into account the particular characteristics of the aquatic environment (such as its dynamics or its light absorption). We therefore discuss in detail the issues encountered in the underwater environment concerning dense matching, calibration, in situ acquisition, data registration and redundancy.

There is ongoing research into building dense correspondence between digital images of objects in the world and estimating the 3D pose of these objects. This is a difficult area to conduct research due to the lack of availability of annotated data. Annotating each pixel is too time-consuming. At the time of this writing, current research has managed to use neural networks to establish a dense pose estimation of human body parts (feet, chest, legs etc.). The aim of this thesis is to investigate if a model can be developed using neural networks to perform dense pose estimation on human feet. The data used in evaluating the model is generated using proprietary tools. Since this thesis is using a custom model and custom dataset, one model will be developed and tested with various experiments to gain an understanding of the different parameters that influence the model’s performance. Experiments showed that a model based on DeepLabV3 is able to achieve a dense pose estimation of feet with a mean error of 1.0cm. The limiting factor for a model’s ability to estimate a dense pose is based on the model’s ability to classify the pixels in an image accurately. It was also shown that discontinuous UV unwrapping greatly reduced the model’s dense pose estimation ability. The results from this thesis should be considered preliminary and need to be repeated multiple times to account for the stochastic nature of training neural networks. / Pågående forskning undersöker hur man kan skapa tät korrespondens mellan digitala bilder av objekt i världen och uppskatta de objektens 3D-pose. Detta är ett svårt område att forska inom på grund av bristen på tillgänglig annoterad data. Att annotera varje pixel är tidskrävande. Vid tiden för detta skrivande har aktuell forskning lyckats använda neurala nätverk för att etablera en tät pose-estimering av mänskliga kroppsdelar (fötter, bröst, ben osv.). Syftet med denna arbete är att undersöka om en modell kan utvecklas med hjälp av neurala nätverk för att utföra dense pose-estimering av mänskliga fötter. Data som används för att utvärdera modellen genereras med hjälp av proprietära verktyg. Eftersom denna arbete använder en anpassad modell och anpassad dataset kommer en modell att utvecklas och testas med olika experiment för att förstå de olika parametrarna som påverkar modellens prestanda. Experiment visade att en modell baserad på DeepLabV3 kan uppnå en dense pose-estimering av fötter med en medelfel på 1,0 cm. Den begränsande faktorn för en modells förmåga att uppskatta en dense pose baseras på modellens förmåga att klassificera pixlarna i en bild korrekt. Det visades också att oregelbunden UV-uppackning avsevärt minskade modellens förmåga att estimera dense pose. Resultaten från denna avhandling bör betraktas som preliminära och behöver upprepas flera gånger för att ta hänsyn till den stokastiska naturen hos träning av neurala nätverk.

Doctor of Philosophy / Department of Mechanical and Nuclear Engineering / William L. Dunn / Dense Plasma Focus (DPF) devices are multi-radiation sources of X rays, neutrons (when working with deuterium), ions, and electrons in pulses typically of a few tens of nanoseconds. The Kansas State University device (KSU-DPF) was commissioned to be used as a radiation source with the Mechanical and Nuclear Engineering Department. The device is operated by a 12.5 µF capacitor which can be charged up to 40 kV storing an energy of 10 kJ. The static inductance and resistance of the device L[subscript]0 and r[subscript]0 were measured to be 91±2 nH and 13±3 mΩ. Experiments have shown that the KSU-DPF device produces 2.45 MeV neutrons with a neutron yield of ~2 × 10^7 and 1.05 × 10^7 n/shots in both axial and radial directions. Ions up to 130 keV were measured using a Faraday Cup. The measured hard X-ray spectrum shows an X-ray emission in the range from 20 to 120 keV with a peak at 50 keV while the average effective energy was estimated, using a step filter method, to be 59±3 keV. The KSU-DPF device was used as a pulsed hard X-ray source for material interrogation studies using the signature-based radiation-scanning (SBRS) technique. The SBRS technique uses template matching to differentiate targets that contain certain types of materials, such as chemical explosives or drugs, from those that do not. Experiments were performed with different materials in cans of three sizes. Nitrogen-rich fertilizers and ammonium nitrate were used as explosive surrogates. Experiments showed 100% sensitivity for all sizes of used samples while 50% specificity for 5 and 1- gallon and 28.57% for quart samples. Simulations using MCNP-5 gave results in good agreement with the experimental results. In the simulations, a larger number of materials, including real explosives were tested. To ensure the feasibility of using the DPF devices for this purpose a second device was simulated and the results were encouraging. Experimental and simulation results indicate that use of DPF devices with simple, room-temperature detectors may provide a way to perform rapid screening for threat materials, especially for places where large number of packages need to be investigated.

Doctor of Philosophy / Department of Mechanical and Nuclear Engineering / William L. Dunn / Dense plasma focus (DPF) devices are pulsed power devices capable of producing short-lived, hot and dense plasmas (~10[superscript]19 cm[superscript]-3) through a fast compression of plasma sheath. A DPF device provides intense bursts of electrons and ion beams, X-rays, and 2.5 MeV neutrons when operated with deuterium through the fusion reaction [superscript]2H(d,n)[superscript]3He. The Kansas State University DPF machine was designed and constructed in early 2010. The device was characterized to determine its performance as a neutron source. The device was shown to produce 5.0x10[superscript]7 neutrons/pulse using a tungsten-copper anode. Such machines have the advantages of being non-radioactive, movable, and producing short pulses (typically tens of nanoseconds), which allows rapid interrogation. The signature-based radiation-scanning (SBRS) method has been used to distinguish targets that contain explosives or explosive surrogates from targets that contain materials called “inert,” meaning they are not explosive-like. Different targets were placed in front of the DPF source at a distance of 45 cm. Four BC-418 plastic scintillators were used to measure the direct neutron yield and the neutrons scattered from various targets; the neutron source and the detectors were shielded with layers of lead, stainless steel, and borated polyethylene to shield against the X-rays and neutrons. One of the plastic scintillators was set at 70[supercript]o and two were set at 110[superscript]o from the line of the neutron beam; a bare [superscript]3He tube was used for detecting scattered thermal neutrons. Twelve metal cans of one-gallon each containing four explosive surrogates and eight inert materials were used as targets. Nine materials in five-gallon cans including three explosive surrogates were also used. The SBRS method indicated a capability to distinguish the explosive surrogates in both experiments, although the five gallon targets gave more accurate results. The MCNP code was used to validate the experimental work and to simulate real explosives. The simulations indicated the possibility to use the time of flight (TOF) technique in future experimental work, and were able to distinguish all the real explosives from the inert materials.

International Telemetering Conference Proceedings / October 25-28, 1999 / Riviera Hotel and Convention Center, Las Vegas, Nevada / This paper presents the future of optical networking via photonic switches as a potential replacement for the existing electronic cross-connects. Although optical amplifiers are now mainstream and wave division multiplexing (WDM) systems are a commercial reality, the industry’s long-term vision is one of the all-optical network. This will require optical switching equipment such as all-optical or “photonic” cross-connect switches that will provide packet switching at an optical layer. Currently, as voice calls or data traffic are routed throughout Range and commercial networks, the information can travel through many fiber-optic segments which are linked together using electronic cross-connects. However, this electronic portion of the network is the bottleneck that is preventing the ideal network from achieving optimal speeds. Information is converted from light into an electronic signal, routed to the next circuit pathway, then converted back into light as it travels to the next network destination. In an all-optical network, the electronics are removed from the equation, eliminating the need to convert the signals and thereby significantly improving network performance and throughput. Removing the electronics improves network reliability and restoration speeds in the event of an outage, provides greater flexibility in network provisioning, and provides a smooth transition when migrating to future optical transmission technologies. Despite the fact that photonic switching remains uncommercialized, it now seems apparent that the core switches in both the public networks and DoD Range networks of the early 21st century will probably carry ATM cells over a photonic switching fabric.

L'étude des plasmas denses et tièdes est un domaine qui suscite aujourd'hui l'intérêt de nombreux groupes de recherche de part son large spectre d'applications. Ce régime de la matière, qui recouvre une densité proche de celle du solide et une température allant de 0.1 à une dizaine d'eV, est encore mal connu et présente une grande complexité de part son caractère partiellement dégénéré (électrons) et partiellement corrélé (ions). Afin d'ex- plorer ce régime, nous proposons de le sonder par spectroscopie d'absorption X près des seuils grâce à une source X ultra-rapide (ps), intense, produite par laser. La spectroscopie d'absorption X près des seuils (XANES, EXAFS) est un diagnostic qui permet l'étude de la structure atomique locale de milieux éventuellement non-cristallins (solides, liquides, plasmas denses et tièdes). Elle nécessite l'utilisation de sources X de large bande spectrale ajustée au seuil d'absorption de l'élément étudié. Ma thèse s'est déroulée en plusieurs étapes. Dans un premier temps, nous nous sommes attachés à développer une source X ultra-brève (quelques picosecondes) créées par laser et adaptée à la spectroscopie d'absorption X de l'aluminium. Nous avons réalisé une telle source X en utilisant le rayonnement de couche M de plasmas d'éléments de Z élevé, produits par une impulsion laser femtoseconde focalisée sur une cible solide : plus particulièrement issu du faisceau de transition 4f −3d. Le spectre d'émission de cette source a été étudié autour du flanc K de l'aluminium, c'est-à-dire dans la gamme de 1.50 à 1.75 keV. Une série d'expériences, consistant en des mesures spectrales et temporelles de l'émission X, a été effectuée avec un laser kHz (5 mJ, 30 fs), focalisé sur différentes cibles solides : Sm, Gd, Dy, Er et Yb. Les résultats ont été comparés avec des simulations Averroès - Transpec (code collisionnel - radiatif hors équilibre thermodynamique local, couplé à un traitement de la physique atomique en superconfigurations). Après optimisation, le rayonnement X étudié présente un spectre large-bande dans la gamme énergétique sou- haitée et une durée d'émission de ∼ 4 ps. Les intensités X atteintes sont de l'ordre de quelques 1e7 photons par tir, par eV et par sr, soit un rendement de 1‰ de l'énergie laser émis dans la gamme 1.50 - 1.75 keV. Dans un second temps, nous avons développé un spectromètre d'absorption X adapté aux caractéristiques des sources X créées par laser. Ce dispositif consiste en la mesure simultanée du spectre transmis par un échantillon d'aluminium et du spectre dit de référence qui consiste à mesurer directement le rayonnement X émis par la source à chaque tir laser. Cette mesure permet ensuite de calculer la transmission et donc l'absorption corrigée des fluctuations tir à tir de la source X. Des spectres d'absorption ont été enregistrés près du seuil K d'un échantillon froid d'aluminium (20°C), en accumulant quelques milliers de tirs laser (soit quelques secondes seulement à 1 kHz). Les structures XANES sont claire- ment identifiées et résolues avec un niveau de bruit inférieur à 1 %. De tels spectres XANES, obtenus sur des installations laser de hautes cadences, ouvrent des perspectives pour la réalisation d'expériences pompe - sonde sur de " petites " installations. Enfin, nous avons réalisé une expérience en collaboration avec le LULI, afin de caracté- riser un échantillon d'aluminium préalablement chauffé de façon isochore par un faisceau de protons issus de l'interaction entre une impulsion laser ultra-intense (∼ 1e19 W/cm2), ultra-courte (fs) et une cible solide d'or. Grâce à cette méthode, l'échantillon est très rapi- dement porté à de relativement hautes températures (jusqu'à une dizaine d'eV). Dans ces conditions, la détente du plasma n'intervenant qu'après quelques dizaines de picosecondes, il est possible de sonder le plasma tiède lorsque sa température est homogène et que sa densité est proche de celle du solide. En accord avec des travaux théoriques, les résultats expérimentaux montrent une disparition des structures XANES lorsque l'alumi- nium atteint une température de l'ordre de 1 eV. On attribue cette disparition à une perte de corrélation ion - ion. En perspective de cette étude, nous nous proposons d'étendre ce travail à l'analyse des plasmas denses et tièdes grâce à des techniques de chauffage diverses comme par exemple le chauffage par laser (fs), par propagation de chocs générés par laser (ns), ou encore par chauffage par rayonnement X ou XUV de type FEL (fs). Il est aussi envisagé d'étendre cette étude à d'autres éléments tels que le Fer.

Ce manuscrit présente une approche novatrice du problème de la mise en correspondance dense de pixel dans les images stéréoscopiques. L'originalité de ce tra- vail repose sur l'utilisation d'un ensemble de techniques de représentation de l'imprécis et de l'incertain permettant de minimiser l'a-priori dans le processus de mise en corres- pondance. Ces techniques s'appuient sur une représentation par une partition floue de l'imprécision de la mesure de luminance induite par l'échantillonnage spatial et sur une classification duale de l'information de luminance associée à chaque pixel. Nous avons associé, à cette représentation, une modélisation de la contrainte épipolaire propre aux capteurs stéréoscopiques étalonnés permettant de prendre en compte l'imprécision des paramètres d'étalonnage. Cette représentation permet d'associer, à chaque pixel d'une des images, un domaine pondéré discret des pixels de l'autre image pouvant lui corres- pondre géométriquement et chromatiquement.Nous proposons ensuite de modifier ces distributions de possibilité en utilisant des contraintes supplémentaires, telles que l'ordre ou l'unicité, pour en augmenter la spé- cificité. Nous utilisons, pour ce faire, un processus itératif et coopératif bipolaire. C'est l'ensemble de cette démarche qui permet de minimiser l'influence de la représentation du problème sur les mesures d'appariements. La validité de notre approche est établie par des expérimentations sur des images stéréo- scopiques de synthèses et réelles dont on connait la verité terrain. Celles-ci permettent de tester le comportement de notre méthode vis-à-vis des variations d'illumination et de disparité.

High-pressure techniques have become increasingly important in the synthesis of ceramic and metallic solids allowing the discovery of new materials with interesting properties. In this research dense solid oxides have been synthesised at high pressures, and structural investigations have been conducted using x-ray and neutron diffraction. The perovskite LaPdO3 has been synthesised at pressures of 6{10GPa. Neutron diffraction studies have been carried out from 7{260K to investigate any structural distortions, particularly related to the possibility of charge order at low temperatures. No reduction in symmetry associated with charge ordering has been observed; the material appears to remain metallic with only one unique Pd site down to 7K. LaPdO3 adopts the GdFeO3-type Pbnm structure. The PdO6 octahedra exhibit a tetragonal distortion throughout the temperature range with a shortening of the apical Pd{O bonds of 2:5% relative to the equatorial bonds. Attempts to prepare analogues of the perovskite containing smaller rare earths have resulted in multi-phase samples, and further RPdO3 perovskites remain inaccessible although there is evidence for a small amount of the perovskite phase in the products of synthesis attempts with neodymium. Three new oxypnictide superconductors, RFeAsO1 xFx (R = Tb, Dy and Ho) have been synthesised at 7{12GPa. The materials are isostructural with other recently discovered iron arsenide superconductors and have Tc's of 52:8 K, 48:5K and 36:2K respectively, demonstrating a downturn in Tc in the series for smaller R. Systematic studies on TbFeAsO0.9F0.1 and HoFeAsO0.9F0.1 show negative values of dTc=dV in contrast to those reported for early R containing materials. Low-temperature neutron diffraction measurements on both materials, and synchrotron studies on HoFeAsO0.9F0.1 reveal no tetragonal to orthorhombic transitions as observed in early R-containing materials with lower doping levels. Magnetic reflections are evident but they are shown to be from R2O3 and RAs impurities with TN's of 5:5K for Tb2O3, 6:5K for HoAs and 1:7K < TN < 4K for Ho2O3. The implications of these results for superconductivity in the iron arsenides are discussed.

Fractured rock formations represent a valuable source of groundwater and can be highly susceptible to contamination by dense, non-aqueous phase liquids (DNAPLs). The goal of this research is to evaluate the effectiveness of three accepted remediation technologies for addressing DNAPL contamination in fractured rock environments. The technologies under investigation in this study are chemical oxidation, bioremediation, and surfactant flushing. Numerical simulations were employed to examine the performance of each of these technologies at the field scale. The numerical model DNAPL3D-RX, a finite difference multiphase flow-dissolution-aqueous transport code that incorporates RT3D for multiple species reactions, was modified to simulate fractured rock environments. A gridding routine was developed to allow the model to accurately capture DNAPL migration in fractures and aqueous phase diffusion gradients in the matrix while retaining overall model efficiency. Reaction kinetics code subroutines were developed for each technology so as to ensure the key processes were accounted for in the simulations. The three remedial approaches were systematically evaluated via simulations in two-dimensional domains characterized by heterogeneous orthogonal fracture networks parameterized to be representative of sandstone, granite, and shale. Each simulation included a DNAPL release at the water table, redistribution to pools and residual, followed by 20 years of ‘ageing’ under ambient gradient conditions. Suites of simulations for each technology examined a variety of operational issues including the influence of DNAPL type and remedial fluid injection protocol. Performance metrics included changes in mass flux exiting, mass destruction in the matrix versus the fractures, and percentage of injected remedial fluid interacting with the target contaminant. The effectiveness of the three remediation technologies covered a wide range; the mass of contaminants destroyed were found to range from 15% to 99.5% of the initial mass present. Effectiveness of each technology was found to depend on a variety of critical factors particular to each approach. For example, in-situ chemical oxidation was found to be limited by the organic material present in the matrix of the rocks, while the efficiency of enhanced bioremediation was found to be related to factors such as the location of indigenous bacteria present in the domain and rate of bioremediation. In the chemical oxidation study, the efficiency of oxidant consumption was observed to be poor across the suite of scenarios, with greater than 90% of the injected permanganate consumed by natural oxidant demand. This study further revealed that the same factors that contributed to forward diffusion of contaminants prior to treatment are critical to this remediation method as they can determine the extent of contaminant destruction during the injection period. Bioremediation in fractured rock was demonstrated to produce relatively good results under robust first-order decay rates and active microorganisms throughout the fractures and matrix. It was demonstrated that under ideal conditions, of the total initial mass present, up to 3/4 could be reduced to ethene, indicating bioremediation may be a promising treatment approach due to the effective penetration of electron donor into the matrix during the treatment period and the ongoing treatment that occurs after injection ceases. However, when indigenous bacteria was assumed to exist only within the fractured walls of sandstone, it was found that under the same conditions, the rate of dechlorination was 200 times less than the Base Case. Since the majority of the mass resided in the matrix, lack of bioremediation in the matrix significantly reduced the effectiveness of treatment. Surfactant treatment with Tween-80 was proven to be a relatively effective technique in enhanced solubilisation of DNAPL from the fractures within the domain. However, by comparing the aqueous and sorbed mass at the start and end of the Treatment stage, it is revealed that surfactant treatment is not efficient in removing these masses that reside within the matrix. Furthermore, DNAPLs identified in dead end vertical fractures were found to remain in the domain by the end of the simulations across all scenarios studied; indicating that the injected surfactant experiences difficulty in accessing DNAPLs entrapped in dead end fractures. Altogether, the results underscore the challenge of restoring fractured rock aquifers due to the field scale limitations on sufficient contact between remedial fluids and in situ contaminants in all but the most ideal circumstances.

No description available.