Cette thèse traite de la reconstruction 3D par stéréophotométrie, qui consiste à utiliser plusieurs photographies d'une scène prises sous le même angle, mais sous différents éclairages. Nous nous intéressons dans un premier temps à des techniques robustes pour l'estimation des normales à la surface, et pour leur intégration en une carte de profondeur. Nous étudions ensuite deux situations où le problème est mal posé : lorsque les éclairages sont inconnus, ou lorsque seuls deux éclairages sont utilisés. La troisième partie est consacrée à l'étude de modèles plus réalistes, à la fois en ce qui concerne les éclairages et la réflectance de la surface. Ces trois premières parties nous amènent aux limites de la formulation classique de la stéréophotométrie : nous introduisons finalement, dans la partie 4, une reformulation variationnelle et différentielle du problème qui permet de dépasser ces limites. / This thesis tackles the photometric stereo problem, a 3D-reconstruction technique consisting in taking several pictures of a scene under different lightings. We first focus on robust techniques for estimating the normals to the surface, and for integrating these normals into a depth map. Then, we study two situations where the problem is ill-posed: when lightings are unknown and when only two images are used. Part 3 is devoted to more realistic models, in terms of lightings and of surface reflectance. These first three parts bring us to the limits of the usual formulation of photometric stereo: we eventually introduce in Part 4 a variational and differential reformulation of this problem which allows us to overcome these limits.

Dans ce travail, les dégradations des diodes électroluminescentes (DEL) ont été étudiées en identifiant et en évaluant leurs signatures électriques et photométriques en vieillissement accéléré sous stress thermique et électrique. Un prototype de banc de test expérimental a été développé et construit spécifiquement pour cette étude ce qui nous a permis de tester 128 échantillons en appliquant différentes conditions de stress thermiques et électriques. Quatre types différents de DEL ont été étudié avec des caractéristiques techniques similaires (température de couleur, courant nominal, mono-puce,...) mais avec des technologies différentes couvrant les principaux acteurs du marché (Cree, Osram, Philips et Seoul Semiconductor). Les échantillons ont d'abord été caractérisés à leur état initial, puis soumis à des conditions de stress électrique (à 350mA ou 1050mA) et thermique (fixé à 50°C). Les mécanismes de défaillance ont été analysés en étudiant l'évolution des signatures électriques et photométriques. Ces caractérisations ont permis d'évaluer et de déterminer l'origine des dégradations à différents niveaux : puce semi-conductrice, interconnexions, phosphore ou encapsulation du dispositif. Les caractérisations électriques nous ont permis d'identifier les mécanismes de dégradation de la puce semi-conductrice et de déterminer la nature des dégradations au niveau du contact ohmique du dispositif (sous fort courant injecté). Les caractérisations photométriques complètent cette étude en évaluant les dégradations associées à l'optique (encapsulation et packaging). / In this work, the degradation of light emitting diodes (LEDs) is studied by identifying and evaluating their aging signature during the stress time. The custom-made experimental test bench is built for realization of the test measurement. Through this experimental test bench, it allows to test a large amount of LED samples and enable to select different temperature condition and different current stress level. There are four different types of LED with similar characteristic in term of their color temperature, IF, VF, power (1W) and as monochip, but different technology coming from Cree, Osram, Philips and Seoul Semiconductor. The devices are firstly characterized their electrical and photometrical characteristic at their initial state, then they are submitted to different current stress condition at low current stress (350mA) and high current stress (1000mA) while the thermal stress is fixed at one temperature (50°C). The study of these devices failure mechanism is archived by using the primary method based on the electrical and photometrical characterization of the devices that allows to evaluate their degradation at different locations of the device components such as semiconductor chip, interconnection and device's package. The electrical characteristic of the device's I-V curve: at low injected current level and reverse bias allow us to identify the degradation characteristic of device's semiconductor chip, at high injected current level allows us to determine the degradation of device's ohmic contact and photometric characteristic allows us to evaluate the degradation of device's package system.

Je présente dans cette thèse de nouvelles contraintes sur la formation et l’évolution des galaxies, en étudiant leur croissance en masse et leur évolution au sein de la toile cosmique depuis l’époque de leur formation jusqu’à maintenant. Pour cela, j’ai créé un catalogue photométrique sur le champ COSMOS. Ce catalogue permet de sonder avec précision l’Univers à haut redshift. J’analyse ce relevé observé à l’aide de relevés virtuels, produits à partir de simulations hydrodynamiques. Ces simulations implémentent nos connaissances sur la formation et l’évolution des galaxies.Dans un premier temps, je montre que l’évolution en redshift des propriétés des galaxies est relativement bien comprise en invoquant des processus qui dépendent essentiellement de la masse, tels que le feedback des étoiles et des AGN. Je souligne également comment nos méthodes observationnelles génèrent des biais dans les propriétés physiques des galaxies calculées à partir de la photométrie.Dans un deuxième temps, je montre comment la dynamique des flots de matière à grande échelle gouverne l’acquisition du moment angulaire des galaxies et halos de matière noire, ce qui implique que certaines propriétés des galaxies sont supposées dépendre de leur environnement anisotrope. J’ai extrait la structure filamentaire du catalogue photométrique que j’ai créé sur le champ COSMOS et j’ai mesuré cette dépendance. Je trouve des gradients de masse et de couleurs dans la direction du filament. Il apparaît que la masse et le moment angulaire des galaxies sont deux quantités interdépendantes et tous deux impactés par leur environnement anisotrope. / I present in this thesis new constraints on galaxy formation and evolution while studying the galaxy mass growth and the co-evolution of the cosmic web and the embedded galaxies, from the epoch of cosmic dawn to today.To do so, I first created a new photometric catalog on the COSMOS field with precise photometric redshifts allowing to probe accurately the high-redshift Universe. I analyze this survey while relying heavily on comparisons with virtual galaxy surveys produced from state-of-the- art cosmological hydrodynamical simulations, which capture all our current knowledge of galaxy formation and evolution.From this comparative analysis, in the first part of my thesis I show that the redshift evolution of galaxy properties is reasonably well understood when invoking mass-dependent processes (AGN and stellar feed- back). I highlight also the effect of simplifying assumptions inherent to our observational methods, which bias the physical properties computed from galaxy photometry.Galaxies and haloes are embedded in the cosmic web, an intricate large-scale structure of walls, filaments and nodes. In the second part of my thesis, I show how galaxies and dark haloes gain their angular momentum from the large-scale flow, implying that some of their properties depend on their anisotropic filamentary environment. I then extract the filamentary structure from the observed photometric catalog and measure the dependence of galaxy properties to the anisotropic environment. I find mass and colour gradients towards the filaments. In turn it emerges that galaxy masses and angular momenta are two dependent quantities impacted by their anisotropic environment.

Cette thèse développe de nouvelles approches pour l'effacement et la révélation de la présence humaine en utilisant des techniques de réalité augmentée. Elle se concentre sur les systèmes projecteur-caméra (ProCams) et leur application dans les «projections intelligentes», où le contenu virtuel projeté est adapté en fonction de l'environnement. Nous nous appuyons sur les travaux antérieurs sur la compensation photométrique pour projeter sur une surface colorée fixe qui permet au spectateur d'observer la même image telle qu'elle apparaîtrait sur une surface blanche. Malgré les différentes méthodes de compensation des couleurs qui ont été proposées au cours de décennie, la compensation appliquée à monde réel avec des couleurs saturées et vives est encore une question ouverte. L'objectif principal de ce travail est la compensation photométrique sur un objet 3D en mouvement en utilisant un ProCam, ce qui est un scénario considérablement plus difficile. Une nouvelle méthode pour la compensation de couleur à l'aide d'une image d'entrée unique est proposée. Elle consiste à estimer la réflectance spectrale de la surface afin de compenser celle-ci en supposant que les réponses du projecteur sont connues ou mesurées précédemment. Cette méthode a été entièrement développée sur GPU pour permettre une compensation en temps réel. Les méthodes antérieures sur la compensation couleur sont discutées et comparées afin d'évaluer la performance de notre technique. L'étalonnage photométrique et géométrique précis d'un ProCam est essentiel pour une compensation précise. Une nouvelle technique de calibration géométrique entre un capteur de profondeur et un ProCam est présentée dans le but de réaliser la projection sur un objet en mouvement. Une calibration photométrique est décrite pour la transformation des valeurs RGB de la caméra et du projecteur (dépendantes du périphérique) vers l'espace couleur CIE XYZ 1931 (indépendantes du périphérique). Le potentiel artistique des techniques de réalité augmentée proposées est en outre exploré à travers l'installation interactive artistique “Gardien du Temple”. La révélation et l'effacement ont toujours été un terrain d'expérimentation commun aux artistes et aux scientifiques. Cette installation audiovisuelle utilise la réalité augmentée pour effacer la présence d'un poème écrit sur un tapis invitant le spectateur à le révéler. Dans de telles applications, la précision et la robustesse des méthodes utilisées sont cruciales. Cette installation artistique a donc permis de tester et d'évaluer les travaux de cette thèse dans un contexte collaboratif et exigeant. / This thesis develops new approaches for human presence concealment and revelation using augmented reality techniques. It focuses on projector-camera systems (ProCams) and their application in “smart projections”, where the projected virtual content is adapted according to the environment. One previously-studied problem is the photometric compensation for projecting on a colored fixed surface that allows the viewer to observe the same image as it would appear on a white surface. Despite the various methods for color compensation that have been proposed the last decade, compensation on a real-world surface with saturated colors and sharp color boundaries is still an open issue. The main objective of this work is the color compensation on a moving 3D object using a ProCam, which is a dramatically more challenging scenario. A novel method for color compensation using a single input frame is proposed. It consists in estimating the spectral reflectance of the surface in order to compensate for it under the assumption that the projector responses are known or previously measured. This method was fully developed on GPU to enable real-time compensation. Previous methods on color compensation are discussed and compared in order to evaluate the performance of our technique. The accurate photometric and geometric calibration of a ProCam is essential for precise compensation. A new geometric calibration technique between a depth sensor and a ProCam is presented in order to make the projection on a moving object feasible. A photometric calibration is described for the transformation of the device-dependent camera and projector values to the device-independent CIE XYZ 1931 color space. The artistic potential of the proposed augmented reality techniques is further explored through the interactive art installation “Gardien du Temple”. Revelation and concealment has always been a common experimentation ground for both artists and scientists. This audio visual installation uses augmented reality to hide the presence of a written poem on a carpet inviting the spectator to reveal it. In such applications, the accuracy and robust performance of the methods employed is crucial and, thus, it offered a challenging ground for testing and evaluation.

In the previous years, 3D imaging has found his way into the world of Egyptology. This lecture will present two case studies where 3D technology is used for the documentation of hieratic inscriptions. The inscriptions, painted in (red) ochre or black paint, were applied on different carriers, and required a different methodology. The Egyptian collection of the Royal Museums of Art and History (RMAH Brussels) contains a large number of small decorated and/or inscribed objects. Some of these objects are currently in a bad condition - any operation carried on them can result in considerable material losses -, making it necessary to document them in such a way that it allows future scholars to study them in detail without handling them. The EES Project therefore aims to create multispectral 3D images of these fragile objects with a multispectral ‘minidome’ acquisition system, based on the already existing system of the multi-light Portable Light Dome (PLD). The texture/colour values on the created 2D+ and 3D models are interactive data based on a recording process with infrared, red, green, blue, and ultraviolet light spectra. Software tools and enhancement filters have been developed which can deal with the different wavelengths in real-time. This leads to an easy and cost-effective methodology which combines multispectral imaging with the actual relief characteristics and properties of the physical object. The system is transportable to any collection or excavation in the field. As a case study, the well-known Brussels “Execration Figurines” (Middle Kingdom, c. 1900 BC) were chosen. These figurines are made of unbaked clay and covered with hieratic texts, listing names of foreign countries and rulers. The study of this type of collections is mostly hampered by the poor state of conservation of the objects, but also by the only partial preservation of the ink traces in visible light. The method has also been applied to other decorated objects of the RMAH collection, such as a Fayoum portrait, ostraca and decorated objects made of stone, wood and ceramics. The final goal will be to publish the newly created multispectral 3D images on Carmentis (www.carmentis.be), the online catalogue of the RMAH collection, making them accessible to scholars all over the world. The second case study presents the quarry inscriptions of the New Kingdom limestone quarries at Dayr Abu Hinnis (Middle Egypt). These gallery quarries contain hundreds of hieratic inscriptions, written on the ceiling. The texts are mainly related to the general administration of the quarry area. In documenting the abundance of ceiling inscriptions and other graffiti, we had to decide upon a practice that would allow not only to capture the \"content\", but also to document the location and orientation of each record. Every inscription can be photographed in detail, but this is insufficient to provide the reader access to vital information concerning the spatial distribution of the inscriptions, which may, for instance, relate to the progress of work. After experimenting with a variety of other methods, we adopted a photogrammetric software for 3D modelling photographs of the quarry ceilings, AGISOFT PHOTOSCAN, which uses structure from motion (SFM) algorithms to create three-dimensional images based on a series of overlapping two-dimensional images. The ultimate goal of this whole labour-intensive process in the quarries is not the creation of pure threedimensional models, but rather to generate an orthophoto of the entire ceiling of a quarry. Based on these images, each graffito could be analysed in context.

Avec près de 4000 exoplanètes connues, le domaine est passé de simplement détecter des exoplanètes à étudier leurs propriétés atmosphériques. Cependant, les spectres en lumières réfléchies de ces objets sont encore mal compris. Les exoplanètes réfléchissent une partie de la lumière qu’elles reçoivent de leur étoile, selon les propriétés de l’atmosphère, ce qui affecte le budget énergétique de la planète. Les Jupiters chaudes, c’est-à-dire des planètes de types Jupiter avec des périodes orbitales très courtes, sont les cibles les plus faciles à observer par spectroscopie des éclipses. L’albédo est une mesure directe de la lumière réfléchie qui peut être mesurée pendant que la planète passe derrière l’étoile hôte. Dans leur cas spécifique, une incohérence apparente, appelée le problème d’albédo des Jupiters chaudes, reste non résolu. Alors que les géantes gazeuses du système solaire ont des albédos de Bond inférieurs aux albédos géométriques, les mesures dans le visible et l’infrarouges pour HD 189733b et HD 209458b indiquent le contraire. Ceci pourrait être expliqué par des albédos géométriques plus élevés à des longueurs d’onde UV/visibles hors de la bande passante de Kepler, mais très peu de mesures existent pour corroborer cela. Ce mémoire présente le spectre de réflexion complet de WASP-43b, incluant 3 mesures d’éclipse obtenues par le HST (290-570 nm) ainsi que 28 obtenues par la mission TESS (600-1000 nm). Lorsque combinées avec les observations Spitzer ou les observations d’éclipse du JWST à venir, ces mesures répondront à des questions-clés concernant la structure et composition atmosphérique de la planète, le budget énergétique global et sa circulation. / With nearly 4000 exoplanets known, the field has evolved from merely detecting exoplanets to actually probing atmospheric properties. However, reflected light spectra from these objects are still not fully understood. Exoplanets reflect a portion of the light that they receive from the star, the amount of which depends on the properties of the atmosphere and in turn affects the energy budget of the planet. Hot Jupiters, i.e. Jupiter-like planets giants with very short orbital periods are the easiest targets amenable to eclipse spectroscopy. Albedo is a direct measure of reflected light that can be measured while the planet eclipses behind the host star. In the specific case of these intriguing planets, an apparent inconsistency, termed as the hot Jupiter Albedo Problem, remains unsolved. While Solar System gas giants show Bond albedos lower than geometric albedos, the measurements from optical and infrared instruments for HD 189733b and HD 209458b show the opposite. This phenomenon has the potential to be explained by higher geometric albedos at UV/optical wavelengths outside the Kepler bandpass, but very few measurements exist to corroborate this. This thesis presents WASP-43b’s full reflection spectrum, including 3 eclipse measurements obtained by the HST (290-570 nm) along with 28 obtained by the TESS mission (600-1000 nm). When combined with the Spitzer or the upcoming JWST’s eclipse observations, these measurements will answer key questions about the planet’s atmospheric composition and structure, global energy budget and circulation.

Shape and attitude of near-Earth objects directly affect the orbit propagation via drag and solar radiation pressure. Obtaining information beyond the object states (position and velocity) is integral to identifying an object. It also enables tracing origin and can improve the orbit accuracy. For objects that have a significant distance to the observer, only non-resolved imaging is available, which does not show any details of the object. So-called non-resolved light curve measurements, i.e. photometric measurements over time can be used to determined the shape of space objects using a two step inversion scheme. It follows the procedure to first determine the Extended Gaussian Image and then going through the shape reconstruction process to retrieve the closed shape even while measurement noise is present. Furthermore, it is also possible to generate high confidence candidates when follow-up observations are provided through a multi-hypotheses process.

The Milky Way galaxy, like all spiral galaxies, is surrounded by a roughly spherical distribution of stars called the halo. The halo was largely formed when the galaxy merged with smaller galaxies. The stellar population of the inner halo is dominated by debris from one major such merging event, called the Gaia-Enceladus-Sausage, and the outer halo population is completely built up by several mergers. To properly understand this accretion history, the halo needs to be investigated out to large distances so that as much substructure as possible can be traced. The substructure is expected to leave an imprint in the halo's metallicity structure. In this thesis, we use a catalogue of intrinsically bright stars, giants, to probe the metallicity structure of the halo to large distances. It contains 205,727 stars that all have photometric metallicities from the Pristine survey and distances derived from isochrone fitting, and reaches down to [Fe/H] = -4.0 dex and out to d = 96.16 kpc. Its purity is 90 % and completeness is 67 %. We calculate the distance errors by Monte Carlo simulations and introduce a new cut in colour that is dependent on metallicity to reduce contaminants in the sample. This introduces a metallicity bias in the sample that we can correct for because we coupled metallicity and colour. The correction is done by computing weights for different metallicity bins. The final catalogue allows us to create metallicity distribution functions of the halo as a function of distance. These show us that as heliocentric distance, the distance from the Galactic centre or the Galactic plane increases, the overall metallicity decreases. At the closest distances, the thick disk metallicity peak at -0.7 dex dominates, but as we move further out this smoothly shifts to -1.3 dex and then to -1.6 dex, representing the inner halo, while a peak at -2.2 dex, representing the outer halo, starts to become visible beyond 6 kpc and dominates the metallicity distribution past 26 kpc. These peaks are remnants of merger events in the halo, with the inner halo peak being due to the massive Gaia-Enceladus-Sausage merger and the outer halo peak being due to the many, low-mass and thus low-metallicity galaxies accreted there. We are able to see signals from the Sagittarius stream and Gaia-Enceladus-Sausage's apocentric pile-ups in the metallicity distribution functions, showing that the halo's metallicity changes with not only distance but also sightline. We also detect a diffuse, very metal-poor cloud in the southern footprint that may be an until now unknown structure. This catalogue and its resulting metallicity distribution functions are thus a suitable addition to literature at the metal-poor and distant end, as well as the faint end where e.g. Gaia mission data is unable to provide metallicities and distances. / Vintergatan är, som alla spiralgalaxer, omringad av en ungefärligen sfärisk distribution av stjärnor som kallas halon. Denna halo bildades när galaxen sammansmälte med mindre galaxer. Stjärnpopulationen i den inre halon domineras av stjärnor som kom in med en sådan, väldigt stor, sammanslagning med en annan galax som eter Gaia-Enceladus-Sausage, och den yttre halons stjärnpopulation har helt och hållet byggts upp av sammanslagningar med flera galaxer. För att kunna kartlägga Vintergatans alla sammanslagningar med andra galaxer behöver vi därför undersöka den yttre halon så långt ut som möjligt, för att spåra så mycket substruktur i halon som uppstod till följd av sammanslagningar som möjligt. Sådan substruktur förväntas göra ett avtryck i halons metallicitetsstruktur. I denna masteruppsats används en katalog av intrinsiskt ljusstarka stjärnor, jättar, för att kartlägga halons metallicitetsstruktur till stora avstånd. Katalogen innehåller 205,727 stjärnor som alla har fotometriska metalliciteter från Pristine-undersökningen samt avstånd från isokronanpassning. Den når [Fe/H] = -4.0 dex och d = 96.16 kpc. Mängden jättar i katalogen är 90 % och katalogen behåller 67 % av alla jättar i ursprungskatalogen. Avståndens osäkerhet beräknas med Monte Carlo-simulationer, och det införs ett nytt klipp med färg som är beroende av metallicitet för att undvika kontamination. Detta inför en metallicitetssnedvridning av katalogen som vi kan korrigera eftersom att vi kopplade ihop metallicitet och färg. Korrektionen sker genom att vi beräknar vikter för olika metallicitetsintervall i katalogen. Den slutgiltiga katalogen låter oss skapa metallicitetsdistributioner för halon som beror på avstånd. Dessa distributioner visar oss att när stjärnors avstånd till solen, från galaxens centrum samt från galaxdisken ökar, så minskar den genomsnittliga metalliciteten. Vid väldigt nära avstånd är distributionerna centrerade runt -0.7 dex som motsvarar den tjocka disken, men när avstånden ökar, flyttas denna topp till -1.3 dex och sedan till -1.6 dex, vilket motsvarar den inre halon, samtidigt som en topp vid -2.2 dex, som motsvarar den yttre halon, framträder bortom 6 kpc och dominerar metallicitetsdistributionen för halon bortom 26 kpc. Dessa toppar är kvarlevor efter sammanslagningar mellan Vintergatan och mindre galaxer, där toppen i den inre halon uppstod p.g.a. Gaia-Enceladus-Sausage och toppen i den yttre halon kommer från de många lågmassiva och därmed metallfattiga dvärggalaxer som assimilerats där. Metallicitetsdistributionerna visar signaler från Sagittariusströmmen och Gaia-Enceladus-Sausages apocentriska hopsamlingar, vilket visar att halons metallicitet inte bara beror på avstånd utan även på observationsvinkel. Katalogen visar spår av ett diffust metalfattigt moln i det södra observationsfönstret som potentiellt är en hittills oupptäckt struktur. Denna katalog och dess resulterande metallicitetsdisitributioner är en viktig addering till existerande litteratur i den metallfattiga och avlägsna regimen, samt i den ljussvaga regimen där t.ex. data från Gaiateleskopet inte kan bidra med metallicitets- eller avståndsvärden.

No description available.

The Milky Way's halo is an approximately spherical distribution of stars surrounding the Galaxy that carries the history of the Milky Way. The outer halo is a Galactic region with long dynamical timescales largely built up by accreted material. Probing its stellar constituents has been historically difficult due to the distances of outer halo stars, making them appear faint. To characterise the distant halo and unravel the history of our galaxy, we thus need to use stars that are intrinsically bright, i.e. giant stars. To draw useful conclusions about the distant halo, these target giants should have metallicity and kinematics information. Therefore a catalogue of distant halo giants with Pristine survey metallicities, Gaia mission data and distances has been created in this work. The cuts used to create this catalogue are made to remove as many dwarf stars as possible and have been tested on a training sample containing spectroscopic metallicities and surface gravities as well as Gaia mission data. Defining giants as being all stars with log(g) < 3.5 dex, we can calculate the purity and completeness of the sample after the cuts have been applied to test which cuts optimise the catalogue. The methods used to cut away the dwarfs are to first plot all stars with positive Gaia parallaxes and fractional parallax uncertainties smaller than 50% in a colour-absolute magnitude diagram and remove all stars from the sample that in this plot populate the main sequence. We then colour-code the colour-apparent magnitude diagram by purity and completeness after this parallax cut has been performed, and select a region in this diagram in which both purity and completeness are maximised, with the final region being (GBP,0 - GRP,0) > 0.8 and G0 < 17.6. The distances to the stars in this region are then computed by comparing their apparent magnitudes to the absolute ones of isochrones. These cuts are then applied to a sample of 6,884,547 stars with Pristine survey and Gaia mission data. The final catalogue is kinematically unbiased and contains 345,303 halo giants. It contains 78% giants and only 4% of giants are erroneously deselected.  With the final sample we are able to probe as deep as 103 kpc into the halo and have created preliminary metallicity distribution functions of different regions of the halo. This sample will be used to further investigate the distant halo metallicity structure and its substructure that was created through merger events.