The progression of technical development has yielded practicable camera systems for the acquisition of so called depth maps, images with depth information. Images and movies with depth information open the door for new types of applications in the area of computer graphics and vision. That implies that they will need to be processed in all increasing volumes. Increased depth image processing puts forth the demand for a standardized data format for the exchange of image data with depth information, both still and animated. Software to convert acquired depth data to such videoformats is highly necessary. This diploma thesis sheds light on many of the issues that come with this new task group. It spans from data acquisition over readily available software for the data encoding to possible future applications. Further, a software architecture fulfilling all of the mentioned demands is presented. The encoder is comprised of a collection of UNIX programs that generate MPEG Z/Alpha, an MPEG2 based video format. MPEG Z/Alpha contains beside MPEG2's standard data streams one extra data stream to store image depth information (and transparency). The decoder suite, called TexMPEG, is a C library for the in-memory decompression of MPEG Z/Alpha. Much effort has been put into video decoder parallelization, and TexMPEG is now capable of decoding multiple video streams, not only in parallel internally, but also with inherent frame synchronization between parallely decoded MPEG videos.

Au cours de leur mise en forme, les alliages de titane destinés à la fabrication de pièces tournantes pour moteurs d'avions suivent un schéma thermomécanique complexe constitué généralement de plusieurs opérations de forgeage. L'étape de forgeage dans le domaine monophasé β joue un rôle important, car elle conditionne la transformation de phases qui s'opère au cours du refroidissement ou du traitement thermique consécutif dans le domaine biphasé α-β, et donc les propriétés mécaniques finales de la pièce. La maîtrise de l'évolution de la microstructure et de la texture cristallographique au cours du forgeage dans le domaine β est donc indispensable pour optimiser l'ensemble du procédé de mise en forme. Pour cela, il est important de bien comprendre les mécanismes de recristallisation dynamique (continue et géométrique) se produisant dans ce type de matériau.Dans cette étude, les effets de la température, de la déformation et de la vitesse de déformation ont été étudiés au moyen d'essais de compression uniaxiale et de torsion à chaud. Les microstructures et textures de déformation ont été caractérisées par diffraction des électrons rétrodiffusés (EBSD). Ces approches expérimentales sont associées à un travail de modélisation, pour lequel le schéma de recristallisation dynamique "continue" (ou CDRX), consistant en une fragmentation progressive des grains β initiaux par la formation de nouveaux joints de grains, a été adapté à partir d'études antérieures. Les paramètres caractérisant les mécanismes élémentaires de la CDRX (écrouissage, restauration dynamique, migration des joints) ont été déterminés par ajustement sur les données expérimentales.Ces travaux ont pour objectif l'élaboration d'un post-processeur métallurgique dédié au forgeage des alliages de titane en phase β. / Titanium alloys used for the manufacture of rotating parts of aircraft engines undergo complex thermomechanical processing schedules, involving usually several forging operations. The forging step in the single -phase domain plays an important role, because it governs the phase transformation that occurs during cooling or subsequent heat treatment in the two-phase α+β range, and therefore the final mechanical properties of the workpiece. Controlling the microstructural evolutions during forging in the β-domain is of prime interest to optimize the entire process of disks manufacture. A full understanding of the (continuous and geometric) dynamic recrystallization mechanisms occurring in this type of material is therefore required.In this study, the effects of temperature, strain and strain rate were investigated by hot uniaxial compression and torsion tests. The microstructures and textures were characterized using backscattering electron diffraction (EBSD). These experimental approaches were associated with the development of a physical model. In the latter, the mechanism of "continuous" dynamic recrystallization (CDRX), consisting of a progressive initial fragmentation of the grains leading to the formation of new grain boundaries, was adapted from former studies. The parameters characterizing the elementary mechanisms of CDRX (strain hardening, dynamic recovery, grain boundary migration) were determined by fitting the experimental data.The final objective of this work is to develop a metallurgical post-processor dedicated to forging operations of titanium alloys in the -phase range.

Dans cette thèse on a étudié par diffraction des rayons X des polytypes à longue période de micas, qui sont des cristaux naturels formés en condition hors équilibre dans un système magmatique. Leur séquence d'empilement peut être considérée comme une combinaison plus ou moins complexe de trois polytypes à courte période 1M, 2M1 et 3T, appelés « structures de base ». L'objectif est d'obtenir les premiers affinements structuraux des ces polytypes avec attention particulière à la distribution des cations dans chaque feuillet, car, d'après la littérature, il existe vraisemblablement une relation entre cette distribution et le mécanisme de formation de ces polytypes. Plus de 120 cristaux ont été analysés, parmi lesquels deux polytypes, dont la période est formée respectivement par 4 et 5 feuillets (indiqués comme 4M3 et 5M3 respectivement en notation de Ramsdell), ont été sélectionnés. Grâce aux résultats de l'affinement, deux interprétations de formation sont possible pour le polytype 4M3, qui prennent en considération le fait que la séquence d'empilement peut-être formée par l'inter-croissance de deux sous-structures : une 3T avec une 1M ou deux sous-structures 2M1 ; dans le dernier cas il s'agit d'un exemple de « cell-twin ». Les résultats de l'affinement structural du polytype 5M3 ont permis de prouver pour la première fois la coexistence des feuillets de type M1 et M2, qui diffèrent pour la distribution cationique dans la couche octaédrique, dans le même polytype. La formation de ces polytype peut-être due à la combinaison de deux sous-structures de base et plusieurs interprétations sont possibles : 3T+1M, 3T+2M1, 1M+2M1, ou deux 1M. Une étude géologique a été effectuée pour reconstruire l'histoire de la rhyodacite contenant les polytypes à longue période. Cette roche s'est formée suite à un mélange de magma, comme indiqué par la présence de textures de déséquilibre de plusieurs minéraux, phénomène qui semble avoir un rôle important dans la formation des ces polytypes, en incrémentant l'activation et les interactions des spirales de croissance, mais aussi les interactions spirale-cristal et cristal-cristal. / We have studied by X-ray diffraction long period polytypes of micas, that are natural crystals formed far from the equilibrium condition in a magmatic system. Their stacking sequence can be considered as a more or less complex combination of three short polytypes 1M, 2M1 and 3T, called "basic structures". The aim was to obtain the first structural refinements of these polytypes, with special attention to the cationic distribution in the octahedral sheet, because, according to the literature, there seems to be a relation between this distribution and the formation of these polytypes. More than 120 crystals have been analysed, and two polytypes have been chosen, whose period is formed respectively by 4 and 5 layers (indicated as 4M3 and 5M3 respectively in Ramsdell notation). The refinement results suggest two interpretations for the formation of the 4M3 polytype, both indicating that the stacking sequence can be formed by the inter-growth of two sub-structures: a 3T with a 1M or two 2M1 sub-structures; the last case is an example of "cell twin". The results of the structural refinement of the 5M3 polytype show for the first time the coexistence of M1 and M2 types of layer, which differ for the cation distribution in the octahedral sheet, in the same polytype. The formation of this polytype can be due to the combination of two basic sub-structures and several interpretations are possible: 3T+1M, 3T+2M1, 1M+2M1, or two1M. A geological study has been done to reconstruct the history of the rhyodacite containing the long period polytypes. This rock is formed after a magma mixing, as showed by the presence of the disequilibrium textures of several minerals; this phenomenon seems to play a fundamental role for the formation of these polytypes, increasing the spiral?s activation and interactions, but also the spiral-crystal and crystal-crystal interactions.

A Nappe Lima Duarte está situada no sudeste do Orógeno Brasília. É constituída por paragnaisses migmatíticos com granada, sillimanita, biotita e muscovita, e ortoquartzitos grossos, com intercalações esparsas de gnaisses calciossilicáticos e de anfibolitos. O Complexo Mantiqueira, infraestrutura alóctone da nappe, ocorre na forma de lascas tectonicamente imbricadas na mesma. É constituído por ortognaisses migmatíticos e polimetamórficos, tipo TTG, com intercalações de rochas metabásicas granulíticas, na forma de enclaves máficos alongados e boudins, geralmente concordantes com a foliação principal. Também ocorrem rochas charnockíticas aparentemente intrusivas nos ortognaisses Mantiqueira, com rochas metabásicas associadas. A associação mineral observada nos paragnaisses (Grt + Bt + Sil + Pl + Rt +Ilm + Qtz ± Ms ± Kfs ± Ky) é relacionada a um metamorfismo progressivo de fácies anfibolito superior, caracterizado por reações de quebra de muscovita e geração de feldspato potássico. As condições de pico bárico obtidas no THERMOCALC para a associação com cianita são de 10 ± 0.6 kbar, a 807 ± 25ºC. O pico térmico de 827 ± 44ºC a 8.2 ± 1.8 kbar, no limite da curva de quebra da dumortierita, foi obtido no THERMOCALC com a associação mineral envolvendo sillimanita. As rochas metabásicas inseridas nos ortognaisses do Complexo Mantiqueira e rochas charnockíticas associadas apresentam a associação Grt-Cpx-Pl-Qtz±Opx+Hbl, que é diagnóstica do fácies granulito de alta pressão. São caracterizadas pela presença de texturas coroníticas progressivas de Grt-Cpx-Pl-Qtz nos contatos entre Opx, Pl e/ou opacos, aparentemente de origem ígnea, que marca a passagem do campo dos granulitos de pressão intermediária para o campo dos granulitos de alta pressão. As condições de pico registradas nos veios constituídos por Grt-Cpx-Pl nos metagabronoritos é de 831.8ºC, a 10 kbar. O granada granulito registra o pico metamórfico a 890 ± 41ºC, a 9.26 ± 1.93 kbar. Cálculos realizados no TWEEQU forneceram condições de equilíbrio de 801ºC, a 9.6 kbar para a associação de fácies granulito. As condições de pico bárico nas rochas charnockíticas são de 14.36 ± 1.9 kbar, a 680ºC, enquanto que as temperaturas máximas registradas são de 885.17ºC, a 10 kbar. Cálculos realizados no THERMOCALC forneceram temperatura de 771 ± 166ºC, a 11.8 ± 2.4 kbar. As rochas metabásicas relacionadas ao Complexo Mantiqueira apresentam baixas concentrações de elementos LILE, possivelmente devido ao empobrecimento destes elementos durante o metamorfismo através de perdas por reações de desidratação. Os dados geoquímicos apontam fontes do tipo E-MORB para grande parte das rochas metabásicas, embora sempre com enriquecimento em ETR maior, o que é sugestivo de fontes enriquecidas. O Grt-cpx anfibolito simplectítico apresenta assinaturas geoquímicas distintas, com enriquecimento maior em elementos LILE e ETRL, o que sugere uma origem a partir de fontes OIB. Os padrões de ETR e diagramas de variação multi-elementares de elementos traço sugerem que as rochas charnockíticas têm fontes relacionadas à ambientes de arco vulcânico. Os paragnaisses, em fácies anfibolito superior a granulito, registram uma trajetória inicial horária, descompressiva ao campo da sillimanita. É distinta da trajetória inicial anti-horária exibida pelas rochas metabásicas e charnockíticas, que registram nas coronas de Grt-Cpx-Pl o metamorfismo progressivo de fácies granulito de alta pressão. Sugere-se que esse relativo aumento de pressão tenha sido condicionado pela colocação dos metassedimentos da Nappe Lima Duarte sobre as rochas do Complexo Mantiqueira, porém no mesmo campo de temperatura. Assim, o avanço da nappe metassedimentar pode ter sido responsável pelo soterramento das rochas metabásicas e charnockíticas relacionadas com o Complexo Mantiqueira, o que justifica a pressão mais elevada nestes litotipos. A etapa de exumação foi compartilhada por ambas, o que é evidenciado nas semelhanças de condições metamórficas durante a trajetória de resfriamento quase isobárico, porém com os litotipos do Complexo Mantiqueira em nível crustal mais profundo. / The Lima Duarte Nappe is located in southeastern Brasília Orogen and is composed by migmatitic paragneisses presenting garnet, sillimanite, biotite and muscovite, and coarse-grained orthoquartzites, with few amphibolite and calc-silicate interlayers. The Mantiqueira Complex occurs as tectonic imbricated lenses in the Lima Duarte Nappe, resembling an allochthon structure. It comprises TTG-type migmatitic and polymetamorphic orthogneisses, presenting granulitic metabasic interlayers, as mafic bands and lenses, as well as boudins, which are often concordant with the main foliation. Charnockitic rocks are apparently intrusive in the Mantiqueira orthogneisses, with associated metabasic rocks. The mineral assemblage observed in paragneisses (Grt + Bt + Sil + Pl + Rt + Ilm + Qtz ± Ms ± Kfs ± Ky) is related to an upper amphibolite facies progressive metamorphism characterized by muscovite breakdown reactions producing potassic feldspar. The peak baric conditions obtained in the THERMOCALC processing software for the assemblage involving kyanite are 10 ± 0.6 kbar and 807 ± 25ºC. The thermal peak of 827 ± 44ºC and 8.2 ± 1.8 kbar obtained in THERMOCALC for the assemblage envolving sillimanite, is placed in the boundary of breakdown curve for dumortierite. The metabasic rocks interlayered in Mantiqueira Complex orthogneisses show the Grt-Cpx-Pl-Qtz±Opx+Hbl assemblage, indicating high pressure granulite facies. They are characterized by the presence of Grt-Cpx-Pl progressive coronitic textures between Opx, Pl and/or opaques boundaries, apparently with an igneous origin, which marks the transitions from intermediate pressure granulites field to high pressure granulite field. The peak conditions recorded in Grt-Cpx-Pl veins in metagabbronorites is 831.8ºC, and 10 kbar. The garnet granulite records the metamorphic peak at 890 ± 41ºC, and 9.26 ± 1.93 kbar. Thermobarometric calculations performed at TWEEQU revealed equilibrium conditions at 801ºC, and 9.6 kbar based on granulite facies mineral assemblage. The peak baric conditions achieved by the charnockitic rocks are 14.36 ± 1.9 kbar, and 680ºC, while maximum temperatures recorded are 885.17ºC, and 10 kbar. Thermobarometric calculations performed at THERMOCALC revealed temperatures of 771 ± 166ºC, and 11.8 ± 2.4 kbar. The metabasic rocks related to Mantiqueira Complex show low concentrations of LILE elements, possibly due to the depletion of these elements during metamorphism in dehydrating reactions. Geochemical data point out to E-MORB type sources for the great majority of metabasic rocks, even though with an REE enrichment, suggesting more enriched sources. The symplectitic Grt-Cpx amphibolite show distinct geochemical signatures, characterized by a greater enrichment in LILE and light-REE elements, suggesting an OIB source for their origin. REE patterns and trace element spidergrams suggest that charnockitic rocks sources are related to a volcanic arc tectonic setting. Paragneisses, in upper amphibolite to granulite facies, recorded an initial clockwise path, decompressing to the sillimanite field. It differs from initial counterclockwise path exhibited by the metabasic and charnockitic rocks, which preserves the progressive high pressure granulite facies metamorphism in Grt-Cpx-Pl coronae. This pressure increase is probally related to the metassediments of the Lima Duarte Nappe, that thrusted over the Mantiqueira Complex rocks, although in the same temperature field. The buried character of metabasic and charnockitic rocks may be caused by the thrust of the metassedimentary nappe, justifying the higher pressure found in these lithotypes. The exhumation phase was shared by both of them, which is confirmed in the metamorphic similarities conditions, as they cooled out together in a near isobaric path, although the Mantiqueira Complex lithotypes were in a deeper crustal level.

Características de texturas atuam como bons descritores de imagens e podem ser empregadas em diversos problemas, como classificação e segmentação. Porém, quando o número de características é muito elevado, o reconhecimento de padrões pode ser prejudicado. A seleção de características contribui para a solução desse problema, podendo ser empregada tanto para redução da dimensionalidade como também para descobrir quais as melhores características de texturas para o tipo de imagem analisada. O objetivo deste trabalho é avaliar métodos ótimos e subótimos de seleção de características em problemas que envolvem texturas de imagens. Os algoritmos de seleção avaliados foram o branch and bound, a busca exaustiva e o sequential oating forward selection (SFFS). As funções critério empregadas na seleção foram a distância de Jeffries-Matusita e a taxa de acerto do classificador de distância mínima (CDM). As características de texturas empregadas nos experimentos foram obtidas com estatísticas de primeira ordem, matrizes de co-ocorrência e filtros de Gabor. Os experimentos realizados foram a classificação de regiôes de uma foto aérea de plantação de eucalipto, a segmentação não-supervisionada de mosaicos de texturas de Brodatz e a segmentação supervisionada de imagens médicas (MRI do cérebro). O branch and bound é um algoritmo ótimo e mais efiiente do que a busca exaustiva na maioria dos casos. Porém, continua sendo um algoritmo lento. Este trabalho apresenta uma nova estratégia para o branch and bound, nomeada floresta, que melhorou significativamente a eficiência do algoritmo. A avaliação dos métodos de seleção de características mostrou que os melhores subconjuntos foram aqueles obtidos com o uso da taxa de acerto do CDM. A busca exaustiva e o branch and bound, mesmo com a estratégia floresta, foram considerados inviáveis devido ao alto tempo de processamento nos casos em que o número de característica é muito grande. O SFFS apresentou os melhores resultados, pois, além de mais rápido, encontrou as soluções ótimas ou próximas das ótimas. Pôde-se concluir também que a precisão no reconhecimento de padrões aumenta com a redução do número de características e que os melhores subconjuntos freqüentemente são formados por características de texturas obtidas com técnicas diferentes / Texture features are eficient image descriptors and can be employed in a wide range of applications, such as classification and segmentation. However, when the number of features is considerably high, pattern recognition tasks may be compromised. Feature selection helps prevent this problem, as it can be used to reduce data dimensionality and reveal features which best characterise images under investigation. This work aims to evaluate optimal and suboptimal feature selection algorithms in the context of textural features extracted from images. Branch and bound, exhaustive search and sequential floating forward selection (SFFS) were the algorithms investigated. The criterion functions employed during selection were the Jeffries-Matusita (JM) distance and the minimum distance classifier (MDC) accuracy rate. Texture features were computed from first-order statistics, co-occurrence matrices and Gabor filters. Three different experiments have been conducted: classification of aerial picture of eucalyptus plantations, unsupervised segmentation of mosaics of Brodatz texture samples and supervised segmentation of MRI images of the brain. The branch and bound is an optimal algorithm and many times more eficient than exhaustive search. But is still time consuming. This work proposed a novel strategy for the branch and bound algorithm, named forest, which has considerably improved its performance. The evaluation of the feature selection methods has revealed that the best feature subsets were those computed by the MDC accuracy rate criterion function. Exhaustive search and branch and bound approaches have been considered unfeasible, due to their high processing times, especially for high dimensional data. This statement holds even for the branch and bound with the forest strategy. The SFFS approach yielded the best results. Not only was it faster, as it also was capable of finding the optimal or nearly optimal solutions. Finally, it has been observed that the precision of pattern recognition tasks increases as the number of features decreases and that the best feature subsets are those which possess features computed from distinct texture feature methods

[pt] A principal motivação para realização deste trabalho é verificar a viabilidade de uso da técnica de mapeamento de textura com relevo em aplicações que exijam interação em tempo real como, por exemplo, jogos eletrônicos. Esta dissertação apresenta uma extensão ao mapeamento de textura com relevo que suporta a representação de efeitos dependentes do ponto de vista e da direção de iluminação, que antes não eram possíveis de serem representados. Além disso, propõe uma forma de paralelizar tal mapeamento entre CPU e GPU (Graphic Processor Unit), utilizando-se para isto da tecnologia de Hyper-Threading. Nesta nova abordagem, cada amostra pertencente a uma textura com relevo é aumentada para incluir três valores escalares representando o vetor normal à superfície reproduzida pela textura. Desta forma, o cálculo de iluminação pode ser efetuado por pixel. A paralelização do processo de mapeamento de textura com relevo demonstra-se favorável uma vez que são obtidos ganhos de até 37% em relação ao tempo de processamento da abordagem convencional. / [en] The main motivation for this work is to verify the feasibility of using relief texture mapping in high-demand real-time applications, such as computer games. This dissertation presents an extension to relief texture mapping that supports the representation of effects that depend of viewpoint and illumination direction, which could not be represented with previous techniques. Moreover, it proposes a way to parallelize relief texture mapping between CPU and GPU, using Hyper-Threading technology. In this new approach, each element that belongs to a relief texture is augmented to include three scalar values representing the normal vector of the reproduced surface. So, the illumination can be calculated per pixel. In the proposed approach, the parallelization of the relief texture mapping represents an acceleration of up to 37% when compared to conventional techniques.

La synthèse de textures est une technique qui génère une texture automatiquement grâce à un algorithme. Cette technique permet de réduire le temps de création des textures et le coût mémoire étant donné que seuls les algorithmes et leurs paramètres ont besoin d'être stockés. Cependant, des difficultés sont souvent rencontrées lors de l'utilisation des textures synthétisées. D'abord, les paramètres de ces textures sont difficiles à manipuler. Ensuite, l'algorithme de synthèse génère souvent les textures sous forme de tableaux de pixels nécessitant beaucoup de mémoire. Pour aborder ces difficultés, nous proposons les approches suivantes : pour améliorer la visualisation de l'espace des textures synthétisées, nous proposons de construire un résumé de cet espace: une seule image statique qui résume, dans un espace limité de pixels, les apparences produites par un synthétiseur donné. De plus, pour améliorer la sélection de paramètres, nous augmentons les curseurs qui contrôlent les paramètres avec des bandes visuelles révélant les changements qui se produisent quand l'utilisateur manipule les curseurs. Pour permettre à l'utilisateur d'interagir de manière interactive avec les résumés visuels, nous nous reposons sur un algorithme de synthèse par patch permettant de générer les textures de façon rapide grâce à une implémentation parallèle sur le processeur graphique. Au lieu de générer le résultat de ce synthétiseur sous forme d'un tableau de pixels, nous représentons le résultat dans une structure compacte et nous utilisons une méthode rapide permettant de lire des pixels directement à partir de cette structure / Texture synthesis is a technique that algorithmically generates textures at rendering time. The automatic synthesis reduces authoring time and memory requirements since only the algorithm and its parameters need to be stored or transferred. However, two difficulties often arise when using texture synthesis: First, the visualization and parameters selection of synthesized textures are difficult. Second, most synthesizers generate textures in a bitmap format leading to high memory usage. To address these difficulties we propose the following approaches: First, to improve the visualization of synthesized textures we propose the idea of a procedural texture preview: A single static image summarizing in a limited pixel space the appearances produced by a given synthesizer. The main challenge is to ensure that most appearances are visible, are allotted a similar pixel area, and are ordered in a smooth manner throughout the preview. Furthermore, to improve parameters selection we augment sliders controlling parameters with visual previews revealing the changes that will be introduced upon manipulation. Second, to allow user interactions with these visual previews we rely on a fast patch-based synthesizer. This synthesizer achieves a high degree of parallelism and is implemented entirely on the GPU. Finally, rather than generating the output of the synthesizer as a bitmap texture we encode the result in a compact representation and allow to decoding texels from this representation during rendering

Trois magmas (haplotonalite synthétique, basalte et dacite de Santorin) ont été juxtaposés et déformés à l’aide d’une presse de type Paterson pour contraindre les conditions de mélange de magmas à haute pression (300 MPa). Ces trois magmas ont été utilisés en conditions sèches ou hydratées et dans une gamme de température comprise entre 600 et 1200°C, permettant d’obtenir plusieurs fractions cristallines et contrastes de viscosité. Les textures de mélange produites lors des expériences dépendent de la fraction cristalline et sont similaires aux textures rencontrées dans la nature. Les textures de mélange mécanique (mingling) produites sont la ségrégation de cristaux depuis leur magma source, la formation d’enclaves par détachement et de filaments par étirement de parcelles de magma. Le mélange chimique (mixing) est illustré par des zones d’interactions comportant une large variété de liquides intermédiaires et la cristallisation de nouvelles phases. Le mélange est produit en l’absence d’une charpente cristalline dans l’un ou l’autre des magmas, si le contraste de viscosité est faible (< 0,3 unité log) et en dessous d’un seuil de viscosité absolue, déterminé entre 107 et 108 Pa.s. Par ailleurs, l’eau joue un rôle important sur les propriétés rhéologiques des magmas, et donc sur leurs capacités de mélanges : elle abaisse le seuil de mélange de près de 200°C entre des magmas saturés en eau, et son exsolution (présence de bulles) entraîne une réduction significative des viscosités promouvant le mélange à des fractions cristallines plus faibles. Les conditions rhéologiques favorables aux mélanges entre le magma d’un réservoir et un magma plus chaud qui le recharge ont été déterminées selon la fraction de magma injecté. Cette fraction est de 0,5 minimum pour la majorité des réservoirs typiques du contexte d’arc. La comparaison avec des systèmes plutoniques et volcaniques exposant des figures de mélange montre que les conditions favorables aux mélanges sont atteintes avec des fractions inférieures. Ceci suggère que le fonctionnement d’un réservoir magmatique est influencé par ses styles et taux de recharge, ainsi que la quantité de volatils. / Three magmas (a synthetic haplotonalite, a basalt and a dacite from Santorini Volcano) have been deformed together at P=300 MPa in a Paterson type apparatus to constrain the mixing capacities between chemically contrasted magmas. Both dry and hydrous conditions were investigated at temperatures varying between 600 and 1200°C to vary magma crystallinity and visc osity contrasts. Mingling and mixing features were produced, which closely resemble those encountered in nature. Mingling features are xenocrysts, enclaves, and melt filaments. Diverse intermediate liquids and newly grown crystals characterize the mixing textures. Both mixing and mingling features were produced after crystal network breakage, at low viscosity contrasts (less than 0.3 log unit) and bulk viscosities lower than 107 – 108 Pa.s. The presence of water considerably affects the rheological properties of magmas and their mixing capacities: water-saturated magmas are able to mix at temperatures 200°C colder than their dry equivalent, and bubbles may influence the viscosity contrast leading to mixing at lower crystal fractions. Rheological conditions favourable to magma mixing between the resident magma in the reservoir and a hot replenishing one were determined according to the injected mafic magma fraction. The minimum fraction of mafic magma for mixing is 0.5 for typical arc magma chambers. The comparison with plutonic and volcanic rocks displaying mixing features shows that mixing may, however, occur at lower fractions. This suggests that reservoir dynamism may depend on the style and the frequency of the injection rate, and the amount of volatiles.

Características de texturas atuam como bons descritores de imagens e podem ser empregadas em diversos problemas, como classificação e segmentação. Porém, quando o número de características é muito elevado, o reconhecimento de padrões pode ser prejudicado. A seleção de características contribui para a solução desse problema, podendo ser empregada tanto para redução da dimensionalidade como também para descobrir quais as melhores características de texturas para o tipo de imagem analisada. O objetivo deste trabalho é avaliar métodos ótimos e subótimos de seleção de características em problemas que envolvem texturas de imagens. Os algoritmos de seleção avaliados foram o branch and bound, a busca exaustiva e o sequential oating forward selection (SFFS). As funções critério empregadas na seleção foram a distância de Jeffries-Matusita e a taxa de acerto do classificador de distância mínima (CDM). As características de texturas empregadas nos experimentos foram obtidas com estatísticas de primeira ordem, matrizes de co-ocorrência e filtros de Gabor. Os experimentos realizados foram a classificação de regiôes de uma foto aérea de plantação de eucalipto, a segmentação não-supervisionada de mosaicos de texturas de Brodatz e a segmentação supervisionada de imagens médicas (MRI do cérebro). O branch and bound é um algoritmo ótimo e mais efiiente do que a busca exaustiva na maioria dos casos. Porém, continua sendo um algoritmo lento. Este trabalho apresenta uma nova estratégia para o branch and bound, nomeada floresta, que melhorou significativamente a eficiência do algoritmo. A avaliação dos métodos de seleção de características mostrou que os melhores subconjuntos foram aqueles obtidos com o uso da taxa de acerto do CDM. A busca exaustiva e o branch and bound, mesmo com a estratégia floresta, foram considerados inviáveis devido ao alto tempo de processamento nos casos em que o número de característica é muito grande. O SFFS apresentou os melhores resultados, pois, além de mais rápido, encontrou as soluções ótimas ou próximas das ótimas. Pôde-se concluir também que a precisão no reconhecimento de padrões aumenta com a redução do número de características e que os melhores subconjuntos freqüentemente são formados por características de texturas obtidas com técnicas diferentes / Texture features are eficient image descriptors and can be employed in a wide range of applications, such as classification and segmentation. However, when the number of features is considerably high, pattern recognition tasks may be compromised. Feature selection helps prevent this problem, as it can be used to reduce data dimensionality and reveal features which best characterise images under investigation. This work aims to evaluate optimal and suboptimal feature selection algorithms in the context of textural features extracted from images. Branch and bound, exhaustive search and sequential floating forward selection (SFFS) were the algorithms investigated. The criterion functions employed during selection were the Jeffries-Matusita (JM) distance and the minimum distance classifier (MDC) accuracy rate. Texture features were computed from first-order statistics, co-occurrence matrices and Gabor filters. Three different experiments have been conducted: classification of aerial picture of eucalyptus plantations, unsupervised segmentation of mosaics of Brodatz texture samples and supervised segmentation of MRI images of the brain. The branch and bound is an optimal algorithm and many times more eficient than exhaustive search. But is still time consuming. This work proposed a novel strategy for the branch and bound algorithm, named forest, which has considerably improved its performance. The evaluation of the feature selection methods has revealed that the best feature subsets were those computed by the MDC accuracy rate criterion function. Exhaustive search and branch and bound approaches have been considered unfeasible, due to their high processing times, especially for high dimensional data. This statement holds even for the branch and bound with the forest strategy. The SFFS approach yielded the best results. Not only was it faster, as it also was capable of finding the optimal or nearly optimal solutions. Finally, it has been observed that the precision of pattern recognition tasks increases as the number of features decreases and that the best feature subsets are those which possess features computed from distinct texture feature methods

La détermination des corrélations existant entre la structure et la microstructure d'un matériau et ses propriétés physiques est primordiale pour son optimisation et dans ce contexte, les techniques de diffraction de rayons X, neutrons ou électrons sont des outils de choix. De plus, dans un souci de miniaturisation constant, l'élaboration de matériaux sous forme de films minces et leur éventuelle nanostructuration leur confèrent des propriétés physiques particulières souvent améliorées par rapport au matériau massif, ce qui constitue ces dernières années un enjeu stratégique et économique majeur en raison de leur potentiel d'applications énorme. <br />Dans cet objectif, de nombreuses techniques de dépôt sous vide existent pour l'élaboration de films minces nanostructurés ou non, et elles sont en général séparées en deux grandes catégories avec les techniques de dépôt en phase vapeur chimique (CVD, PECVD, MOCVD...) ou de dépôt physique (évaporation, pulvérisation, ablation laser...). Parmi ces dernières, la technique de pulvérisation magnétron est une des méthodes de dépôt la moins onéreuse, facilement transposable à un niveau industriel et permettant aussi bien de déposer des couches que des multicouches conductrices ou isolantes de bonne qualité cristalline. De plus, la possibilité de l'utilisation d'un gaz réactif tel que l'oxygène ou l'hydrogène est un atout supplémentaire comme nous le verrons dans cette synthèse. <br />L'élaboration de matériaux polyphasés, massifs ou sous forme de films minces, aux propriétés physiques spécifiques très anisotropes a relancé également ces dernières années un intérêt croissant pour l'analyse texturale, puisque l'anisotropie des propriétés physiques macroscopiques d'un échantillon polycristallin ne peut être observée que si ses cristallites sont préférentiellement orientés. C'est le cas notamment des matériaux magnétiques, conducteurs ioniques, ferro- et piézo-électriques et des matériaux à transparence optique. L'analyse quantitative des orientations préférentielles (QTA) par diffraction des rayons X, méthode d'investigation non destructive, joue donc un rôle essentiel dans l'optimisation de ces propriétés anisotropes pour comprendre les corrélations existant entre les tenseurs microscopiques et les propriétés macroscopiques, ainsi que pour déterminer les modes de croissance (épitaxie) et optimiser les conditions d'élaboration. <br />De plus, l'ajout des informations de QTA devient indispensable afin de repousser certaines limites dans des techniques de caractérisation structurale basées sur le rayonnement X, par exemple en EXAFS polarisé, et en couplant les analyses de Rietveld et de texture avec l'analyse dite "combinée" structure/texture/contraintes. En effet, dans un échantillon texturé, si on ne connaît pas exactement sa structure, on ne peut pas en extraire sa texture et vice-versa. De même, la détermination des contraintes dans un échantillon texturé est délicate et une approche par une analyse globale du profil (dite analyse combinée) est souvent nécessaire, encore plus si les échantillons sont polyphasés et de basse symétrie cristalline.<br />Mes activités de recherche jusqu'à aujourd'hui et présentées ci-après visent à établir lorsque cela est possible les corrélations existant entre la structure et la microstructure, déterminées essentiellement par les techniques de diffraction, et les propriétés physiques de matériaux massifs magnétiques ou sous forme de films minces (conducteurs ioniques, ferroélectriques, à transparence optique) en vue de leur optimisation. <br />Ces activités se sont orientées suivant deux axes :<br />• d'une part (Chapitre 1) avec l'analyse quantitative de texture classique et en utilisant principalement la technique de diffraction de rayons X, pour l'optimisation de matériaux massifs intermétalliques ferromagnétiques texturés sous champ magnétique (et étudiés dans le cadre de ma thèse 1996-1999), et de films minces texturés conducteurs ioniques (La,Li)TiO3 ou semi-conducteurs d'InN.<br />• d'autre part (Chapitre 2), avec l'optimisation de films minces texturés en utilisant une approche via l'analyse combinée des spectres de diffraction X. Le recours à une analyse globale du profil pour l'étude des films minces ferroélectriques texturés (Pb,Ca)TiO3 et de TiCx et TiNy, a été nécessaire du fait de recouvrements importants de pics inter-phases. Cette approche dans le cadre de l'optimisation de films minces nanostructurés semi-conducteurs de Si, SiC, Cr2+:ZnSe étudiés au Laboratoire SIFCOM, a été motivée par la présence d'orientations préférentielles et de nanocristallites de tailles anisotropes, ainsi que par la nécessité de lever l'ambiguïté sur la structure du SiC (présence de plusieurs polytypes).<br /><br />Depuis que j'ai rejoint le Laboratoire SIFCOM (septembre 2002), outre des activités ponctuelles de caractérisations structurales par diffraction de rayons X sur des films minces semi-conducteurs texturés de Si et d'InN, mon activité de recherche repose essentiellement sur l'élaboration par pulvérisation magnétron et l'optimisation des propriétés optiques de films texturés de SiC et, depuis février 2005 de films de Cr2+:ZnSe, nouveau sujet de recherche que je développe au laboratoire.