The plant leaf is one of the most challenging natural objects to be realistically depicted by computer graphics due to its complex morphological and optical characteristics. Although many studies have been done on plant modeling, previous research on leaf modeling required for close-up realistic plant images is very rare. In this thesis, a novel method for modeling of the leaf shape based on the leaf venation is presented. As the first step of the method, the leaf domain is defined by the enclosure of the leaf boundary. Second, the leaf venation is interactively modeled as a hierarchical skeleton based on the actual leaf image. Third, the leaf domain is triangulated with the skeleton as constraints. The skeleton is articulated with nodes on the skeleton. Fourth, the skeleton is interactively transformed to a specific shape. A user can manipulate the skeleton using two methods which are complementary to each other: one controls individual joints on the skeleton while the other controls the skeleton through an intermediate spline curve. Finally, the leaf blade shape is deformed to conform to the skeleton by interpolation. An interactive modeler was developed to help a user to model a leaf shape interactively and several leaves were modeled by the interactive modeler. The ray-traced rendering images demonstrate that the proposed method is effective in the leaf shape modeling.

The paper deals with a combination of the Fourier-finite-element method with the Nitsche-finite-element method (as a mortar method). The approach is applied to the Dirichlet problem of the Poisson equation in three-dimensional axisymmetric domains $\widehat\Omega$ with non-axisymmetric data. The approximating Fourier method yields a splitting of the 3D-problem into 2D-problems. For solving the 2D-problems on the meridian plane $\Omega_a$, the Nitsche-finite-element method with non-matching meshes is applied. Some important properties of the approximation scheme are derived and the rate of convergence in some $H^1$-like norm is proved to be of the type ${\mathcal O}(h+N^{-1})$ ($h$: mesh size on $\Omega_a$, $N$: length of the Fourier sum) in case of a regular solution of the boundary value problem. Finally, some numerical results are presented.

The paper deals with a combination of the Fourier method with the Nitsche-finite-element method (as a mortar method). The approach is applied to the Dirichlet problem of the Poisson equation in threedimensional axisymmetric domains with reentrant edges generating singularities. The approximating Fourier method yields a splitting of the 3D problem into 2D problems on the meridian plane of the given domain. For solving the 2D problems bearing corner singularities, the Nitsche finite-element method with non-matching meshes and mesh grading near reentrant corners is applied. Using the explicit representation of singular functions, the rate of convergence of the Fourier-Nitsche-mortaring is estimated in some $H^1$-like norm as well as in the $L_2$-norm. Finally, some numerical results are presented.

In this work the h-adaptive FE strategy for solving 2D problems of fracture mechanics is developed and applicated. The stress indicator employed in the adaptive analysis is able to capture high gradients of stress in the vicinity of the defect. Performance of this technique is checked against the defects of different geometry. Finally the above-mentioned technique exposes its suitability for solving fracture problems.

This summary of doctoral dissertation deal with modeling of fracture in various 2D structures. The standard experimental specimens with defects, which are used in evaluation of fracture parameters, are investigated numerically in this work. The original FE generation technique based on the adaptive approach is developed for the 2D structures with defects. The stress indicator employed in the adaptive analysis is able to capture high gradients of stress in the vicinity of the defect. Performance of this technique is checked against the defects of different geometry. Finally the above-mentioned technique exposes its suitability for solving fracture problems.

Triangle mesh models of images are studied. Through exploration, a computational framework for mesh generation based on data-dependent triangulations (DDTs) and two specific mesh-generation methods derived from this framework are proposed. In earlier work, Yang et al. proposed a highly-effective technique for generating triangle-mesh models of images, known as the error diffusion (ED) method. Unfortunately, the ED method, which chooses triangulation connectivity via a Delaunay triangulation, typically yields triangulations in which many (triangulation) edges crosscut image edges (i.e., discontinuities in the image), leading to increased approximation error. In this thesis, we propose a computational framework for mesh generation that modifies the ED method to use DDTs in conjunction with the Lawson local optimization procedure (LOP) and has several free parameters. Based on experimentation, we recommend two particular choices for these parameters, yielding two specific mesh-generation methods, known as MED1 and MED2, which make different trade offs between approximation quality and computational cost. Through the use of DDTs and the LOP, triangulation connectivity can be chosen optimally so as to minimize approximation error. As part of our work, two novel optimality criteria for the LOP are proposed, both of which are shown to outperform other well known criteria from the literature. Through experimental results, our MED1 and MED2 methods are shown to yield image approximations of substantially higher quality than those obtained with the ED method, at a relatively modest computational cost. For example, in terms of peak-signal-to-noise ratio, our MED1 and MED2 methods outperform the ED method, on average, by 3.26 and 3.81 dB, respectively. / Graduate

Cette thèse porte sur le développement de méthodes pour la reconstruction de formes 3D à l’aide de capteurs inertiels et magnétiques. Lorsqu’ils sont placés sur une forme, ces capteurs fournissent des orientations locales de surface mais leur position absolue dans l’espace 3D est inconnue. Les dispositifs que nous considérons dans cette thèse produisent des orientations locales de surface le long d’un réseau de courbes. Reconstruire des formes 3D à l’aide de telles données pose trois types de défis. Tout d’abord, les mesures des capteurs sont bruitées et incohérentes. Deuxièmement, comme les positions sont inconnues, le réseau de courbes acquis doit être reconstruit à partir des orientations. Enfin, une fois le réseau de courbes reconstruit, il est nécessaire de calculer une surface lisse interpolant ce réseau de courbes et les orientations associées. Pour relever ces défis, on formule les différentes étapes de reconstruction comme un ensemble de problèmes d’optimisation. En utilisant des représentations discrètes, ces problèmes sont résolus efficacement et interactivement.Nous présentons deux contributions principales. Tout d’abord, nous introduisons une méthode produisant un réseau de courbes lisses et cohérentes en utilisant les mesures d’orientation et de distance, ainsi qu'un ensemble de contraintes topologiques fournies par l’utilisateur. Notre méthode se base notamment sur une procédure de lissage des orientations motivée par un principe simple: les positions et les normales des courbes doivent coïncider en chaque intersection d'un réseau.Une fois le réseau de courbes reconstruit, nous proposons une méthode permettant de calculer une surface lisse interpolant ce réseau de courbes, ainsi que les orientations associées. Cette méthode a trois étapes. Tout d’abord grâce aux orientations, les cycles de courbes entourant les patchs surfaciques sont déterminés sans ambiguïté. Ensuite les orientations connues le long des courbes sont propagées à travers le maillage initial et utilisées pour estimer la courbure moyenne. Enfin le maillage final est calculé par une méthode basée sur le Laplacien et utilisant l’information de courbure. Les orientations connues sur le réseau de courbes permettent d’obtenir des maillages lisses et de diminuer les erreurs de reconstruction.Les approches précédentes utilisaient des dispositifs statiques placés le long d’un réseau de connectivité fixe entre les capteurs (ruban, grille). Nous explorons dans cette thèse une nouvelle configuration dynamique, consistant à déplacer un dispositif ponctuel sur la surface. En conséquence, il est possible d’acquérir des données le long d’un réseau arbitraire de courbes lisses sur une surface. Les méthodes proposées dans cette thèse ont été testées sur des données réelles acquises avec ces dispositifs mobiles. Des surfaces physiques fabriquées à partir de modèles numériques nous ont permis de faire une évaluation quantitative en calculant l’erreur de reconstruction entre la vraie surface et notre modèle reconstruit. Même pour des formes complexes, l’erreur moyenne reste autour de 1%. / This thesis presents a complete framework for 3D shape reconstruction using inertial and magnetic sensors. When placed onto a shape, these sensors provide local surface orientations along a curve network on the shape, but their absolute position in the world space is unknown. The challenges with this type of 3D acquisition are threefold. First, sensor measurements are noisy and inconsistent. Second, since positions are unknown, the acquired curve network has to be reconstructed from orientations. Finally, the smooth surface needs to be inferred from a collection of curves with normals. To compute the shape from measured data, our main insight is to formulate the reconstruction as a set of optimization problems. Using discrete representations, these optimization problems are resolved efficiently and at interactive time rates.We present two main contributions. First, we introduce a novel method for creating well-connected networks with cell-complex topology using only orientation and distance measurements and a set of user-defined constraints. By working directly with orientations, our method robustly resolves problems arising from data inconsistency and sensor noise. Our approach is driven by a simple principle mostly overlooked in previous works: at each intersection in a curve network, the positions and the normals of two intersecting curves have to coincide.Second, we address the problem of surfacing a closed 3D curve network with given surface normals. Thanks to the normal vector input, the patch-finding problem can be solved unambiguously and an initial piecewise smooth triangle mesh is computed. The input normals are propagated throughout the mesh. Together with the initial mesh, the propagated normals are used to estimate mean curvature vectors. We then compute the final mesh by combining the standard Laplacian-based variational methods with the curvature information extracted from the input normals. The normal input increases shape fidelity and allows to achieve globally smooth and visually pleasing shapes.Previous approaches used static devices placed along a network with fixed connectivity between the sensors (ribbon, grid). We explore a new dynamic setup, which used a single mobile node of sensors. As a consequence, a dense set of data can be acquired along an arbitrary smooth curve network on a surface.The proposed framework was tested on real-world data acquired using two devices equipped with mobile sensors. A quantitative evaluation was performed by computing the error of reconstruction for fabricated surfaces with known ground truth. Even for complex shapes, the mean error remains around 1%.

[pt] No presente trabalho apresentamos um estudo de curvas silhuetas discretas sobre alguns tipos particulares de malhas, com o objetivo de avaliar propriedades dessas curvas. Nosso objeto de estudo são malhas quadrangulares, ou seja, onde todas as faces sejam quadriláteros e também sejam planares. Em particular dois tipos de malhas são discutidas: circular e cônica. Essas malhas são particularmente interessantes em arquitetura para modelagem de estrutura de vidros. A geração das malhas é feita aplicando-se um processo de otimização e em seguida, sobre essas malhas, definimos curvas discretas como candidatas a silhuetas e buscamos medidas de qualidade para essas curvas. / [en] In this work we study discrete silhouette curves on Planar Quad meshes (PQ meshes), with the objective of evaluate some properties of these curves. PQ meshes correspond to planar quadrilaterals meshes, and our interest is focused particularly on two kinds of meshes: Conical and Circular. They are interesting in architecture for design with glass structures. An optimization process is applied for the mesh generation and we follow defining discrete curves on the meshes to obtain silhouette and to measure their quality.

In this thesis, we propose a general computational framework for generating mesh models of single-component (e.g., grayscale) and multi-component (e.g., RGB color) images. This framework builds on ideas from the previously-proposed GPRFSED method for single-component images to produce a framework that can handle images with any arbitrary number of components. The key ideas embodied in our framework are Floyd-Steinberg error diffusion and greedy-point removal. Our framework has several free parameters and the effect of the choices of these parameters is studied. Based on experimentation, we recommend two specific sets of parameter choices, yielding two highly effective single/multi-component mesh-generation methods, known as MED and MGPRFS. These two methods make different trade offs between mesh quality and computational cost. The MGPRFS method is able to produce high quality meshes at a reasonable computational cost, while the MED method trades off some mesh quality for a reduction in computational cost relative to the MGPRFS method. To evaluate the performance of our proposed methods, we compared them to three highly-effective previously-proposed single-component mesh generators for both grayscale and color images. In particular, our evaluation considered the following previously-proposed methods: the error diffusion (ED) method of Yang et al., the greedy-point-removal from-subset (GPRFSED) method of Adams, and the greedy-point removal (GPR) method of Demaret and Iske. Since these methods cannot directly handle color images, color images were handled through conversion to grayscale as a preprocessing step, and then as a postprocessing step after mesh generation, the grayscale sample values in the generated mesh were replaced by their corresponding color values. These color-capable versions of ED, GPRFSED, and GPR are henceforth referred to as CED, CGPRFSED, and CGPR, respectively. Experimental results show that our MGPRFS method yields meshes of higher quality than the CGPRFSED and GPRFSED methods by up to 7.05 dB and 2.88 dB respectively, with nearly the same computational cost. Moreover, the MGPRFS method outperforms the CGPR and GPR methods in mesh quality by up to 7.08 dB and 0.42 dB respectively, with about 5 to 40 times less computational cost. Lastly, our MED method yields meshes of higher quality than the CED and ED methods by up to 7.08 and 4.72 dB respectively, where all three of these methods have a similar computational cost. / Graduate

Nos travaux concernent la mise en oeuvre d’une méthode multiéchelle pour faciliter la simulation numérique de structures complexes, appliquée à la modélisation de composants aéronautiques (notamment pour les pièces tournantes de turboréacteur et des structures composites stratifiées). Ces développements sont basés autour de la méthode Arlequin qui permet d’enrichir des modélisations numériques, à l’aide de patchs, autour de zones d’intérêt où des phénomènes complexes se produisent. Cette méthode est mise en oeuvre dans un cadre général permettant la superposition de maillages incompatibles au sein du code de calcul Z-set{Zébulon, en utilisant une formulation optimale des opérateurs de couplage. La précision et la robustesse de cette approche ont été évaluées sur différents problèmes numériques. Afin d’accroître les performances de la méthode Arlequin, un solveur spécifique basé sur les techniques de décomposition de domaine a été développé pour bénéficier des capacités de calcul offertes par les machines à architectures parallèles. Ces performances ont été évaluées sur différents cas tests académiques et quasi-industriels. Enfin, ces développements ont été appliqué à la simulation de problèmes de structures composites stratifiées. / Our work concerns the implementation of a method for convenient multiscale numerical simulation of complex structures, applied to the modeling of aircraft components (including rotating parts made of jet engine from laminate composite structures). These developments are based on the Arlequin method which allows to enrich numerical modeling, using patches around areas of interest where complex phenomena occur. This method is implemented in a general framework in order to link made of incompatible meshes in the Z-set{Zébulon finite element code, using an optimal formulation of the coupling operators. The accuracy and robustness of this approach were evaluated on various numerical problems. To increase the performance of the Arlequin method, a specific solver based on domain decomposition techniques has been developed to take advantage of computing capabilities offered by parallel machine architectures. Its performance has been evaluated on different numerical assessments from academic to industrial tests. Finally, these developments have been applied to the simulation of problems made of laminate composite structures.