In this thesis we address the problem of digitizing the motion of three-dimensional shapes that move and deform in time. These shapes are observed from several points of view with cameras that record the scene's evolution as videos. Using available reconstruction methods, these videos can be converted into a sequence of three-dimensional snapshots that capture the appearance and shape of the objects in the scene. The focus of this thesis is to complement appearance and shape with information on the motion and deformation of objects. In other words, we want to measure the trajectory of every point on the observed surfaces. This is a challenging problem because the captured videos are only sequences of images, and the reconstructed shapes are built independently from each other. While the human brain excels at recreating the illusion of motion from these snapshots, using them to automatically measure motion is still largely an open problem. The majority of prior works on the subject has focused on tracking the performance of one human actor, and used the strong prior knowledge on the articulated nature of human motion to handle the ambiguity and noise inherent to visual data. In contrast, the presented developments consist of generic methods that allow to digitize scenes involving several humans and deformable objects of arbitrary nature. To perform surface tracking as generically as possible, we formulate the problem as the geometric registration of surfaces and deform a reference mesh to fit a sequence of independently reconstructed meshes. We introduce a set of algorithms and numerical tools that integrate into a pipeline whose output is an animated mesh. Our first contribution consists of a generic mesh deformation model and numerical optimization framework that divides the tracked surface into a collection of patches, organizes these patches in a deformation graph and emulates elastic behavior with respect to the reference pose. As a second contribution, we present a probabilistic formulation of deformable surface registration that embeds the inference in an Expectation-Maximization framework that explicitly accounts for the noise and in the acquisition. As a third contribution, we look at how prior knowledge can be used when tracking articulated objects, and compare different deformation model with skeletal-based tracking. The studies reported by this thesis are supported by extensive experiments on various 4D datasets. They show that in spite of weaker assumption on the nature of the tracked objects, the presented ideas allow to process complex scenes involving several arbitrary objects, while robustly handling missing data and relatively large reconstruction artifacts.

Submitted by Johnny Rodrigues (johnnyrodrigues@ufcg.edu.br) on 2018-07-10T17:54:50Z No. of bitstreams: 1 JAMILSON RAMOS CAMPOS - DISSERTAÇÃO PPGMAT 2006..pdf: 6748825 bytes, checksum: 592fc0c2a05c38766aae5d672c3c4708 (MD5) / Made available in DSpace on 2018-07-10T17:54:50Z (GMT). No. of bitstreams: 1 JAMILSON RAMOS CAMPOS - DISSERTAÇÃO PPGMAT 2006..pdf: 6748825 bytes, checksum: 592fc0c2a05c38766aae5d672c3c4708 (MD5) Previous issue date: 2006-12 / Este trabalho apresenta um estudo de dois modelos computacionais deformáveis aplicados à simulação de tecidos, ambos modelos de mecânica de partículas fisicamente embasados, contemplando algoritmos, um para cada modelo, para o tratamento de colisões. Estudamos um modelos de malha retangular (clássico e pouco robusto) e um baseado em malha triangular (moderno e robusto) através da implementação, simulações e uma análise qualitativa simples entre os resultados visuais obtidos com ambos. Nenhum destes modelos apresenta relações entre deformações tangenciais e normais e portanto, não geram rugas e/ou dobras espontaneamente. Para torná-los mais realísticos, em nossa implementação, propomos o uso de uma força de acoplamento entre as deformações tangenciais normais. / This work presents a study of two deformable computational models applied to the simulation of cloths, both physically based models of particle mechanics, contemplating one per model, collision treatment algorithms. We study a rectangular grid model (classic and not very robust one) and a triangular mesh based model (modern and robust one) throught implementation, simulations and a simple qualitative analysis between visual results reached with them. Neither of these models presents a relationship with tangent and normal deformations therefore don't build folds and/or wrinkles spontaneously. to turn then more realistic, in our implementation, we propose to make use of a coupling force between tangent and normal deformations.

Cette thèse traite du suivi temporel de surfaces déformables. Ces surfaces sont observées depuis plusieurs points de vue par des caméras qui capturent l'évolution de la scène et l'enregistrent sous la forme de vidéos. Du fait des progrès récents en reconstruction multi-vue, cet ensemble de vidéos peut être converti en une série de clichés tridimensionnels qui capturent l'apparence et la forme des objets dans la scène. Le problème au coeur des travaux rapportés par cette thèse est de complémenter les informations d'apparence et de forme avec des informations sur les mouvements et les déformations des objets. En d'autres mots, il s'agit de mesurer la trajectoire de chacun des points sur les surfaces observées. Ceci est un problème difficile car les vidéos capturées ne sont que des séquences d'images, et car les formes reconstruites à chaque instant le sont indépendemment les unes des autres. Si le cerveau humain excelle à recréer l'illusion de mouvement à partir de ces clichés, leur utilisation pour la mesure automatisée du mouvement reste une question largement ouverte. La majorité des précédents travaux sur le sujet se sont focalisés sur la capture du mouvement humain et ont bénéficié de la nature articulée de ce mouvement qui pouvait être utilisé comme a-priori dans les calculs. La spécificité des développements présentés ici réside dans la généricité des méthodes qui permettent de capturer des scènes dynamiques plus complexes contenant plusieurs acteurs et différents objets déformables de nature inconnue a priori. Pour suivre les surfaces de la façon la plus générique possible, nous formulons le problème comme celui de l'alignement géométrique de surfaces, et déformons un maillage de référence pour l'aligner avec les maillages indépendemment reconstruits de la séquence. Nous présentons un ensemble d'algorithmes et d'outils numériques intégrés dans une chaîne de traitements dont le résultat est un maillage animé. Notre première contribution est une méthode de déformation de maillage qui divise la surface en une collection de morceaux élémentaires de surfaces que nous nommons patches. Ces patches sont organisés dans un graphe de déformation, et une force est appliquée sur cette structure pour émuler une déformation élastique par rapport à la pose de référence. Comme seconde contribution, nous présentons une formulation probabiliste de l'alignement de surfaces déformables qui modélise explicitement le bruit dans le processus d'acquisition. Pour finir, nous étudions dans quelle mesure les a-prioris sur la nature articulée du mouvement peuvent aider, et comparons différents modèles de déformation à une méthode de suivi de squelette. Les développements rapportés par cette thèse sont validés par de nombreuses expériences sur une variété de séquences. Ces résultats montrent qu'en dépit d'a-prioris moins forts sur les surfaces suivies, les idées présentées permettent de traiter des scènes complexes contenant de multiples objets tout en se comportant de façon robuste vis-a-vis de données fragmentaires et d'erreurs de reconstruction. / In this thesis we address the problem of digitizing the motion of three-dimensional shapes that move and deform in time. These shapes are observed from several points of view with cameras that record the scene's evolution as videos. Using available reconstruction methods, these videos can be converted into a sequence of three-dimensional snapshots that capture the appearance and shape of the objects in the scene. The focus of this thesis is to complement appearance and shape with information on the motion and deformation of objects. In other words, we want to measure the trajectory of every point on the observed surfaces. This is a challenging problem because the captured videos are only sequences of images, and the reconstructed shapes are built independently from each other. While the human brain excels at recreating the illusion of motion from these snapshots, using them to automatically measure motion is still largely an open problem. The majority of prior works on the subject has focused on tracking the performance of one human actor, and used the strong prior knowledge on the articulated nature of human motion to handle the ambiguity and noise inherent to visual data. In contrast, the presented developments consist of generic methods that allow to digitize scenes involving several humans and deformable objects of arbitrary nature. To perform surface tracking as generically as possible, we formulate the problem as the geometric registration of surfaces and deform a reference mesh to fit a sequence of independently reconstructed meshes. We introduce a set of algorithms and numerical tools that integrate into a pipeline whose output is an animated mesh. Our first contribution consists of a generic mesh deformation model and numerical optimization framework that divides the tracked surface into a collection of patches, organizes these patches in a deformation graph and emulates elastic behavior with respect to the reference pose. As a second contribution, we present a probabilistic formulation of deformable surface registration that embeds the inference in an Expectation-Maximization framework that explicitly accounts for the noise and in the acquisition. As a third contribution, we look at how prior knowledge can be used when tracking articulated objects, and compare different deformation model with skeletal-based tracking. The studies reported by this thesis are supported by extensive experiments on various 4D datasets. They show that in spite of weaker assumption on the nature of the tracked objects, the presented ideas allow to process complex scenes involving several arbitrary objects, while robustly handling missing data and relatively large reconstruction artifacts.

Devido à variação na qualidade e ao ruído nas imagens médicas, a aplicação de técnicas tradicionais de segmentação é geralmente ineficiente. Nesse sentido, apresenta-se um novo algoritmo a partir de duas técnicas: o modelo de contornos deformáveis por fluxo do vetor gradiente (GVF deformable contours) e a técnica de espaço de escalas utilizando o processo de difusão. Assim, foi realizada uma revisão bibliográfica dos modelos que trabalham com os contornos deformáveis, os quais foram classificados em modelos paramétricos e geométricos. Entre os modelos paramétricos foi escolhido o modelo de contornos deformáveis por fluxo do vetor gradiente (GVF). Esta aproximação oferece precisão na representação de estruturas biológicas não observada em outros modelos. Desta forma, o algoritmo apresentado mapeia as bordas (edge map) e aperfeiçoa a condução da deformação utilizando uma técnica baseada em operações recursivas. Com este cálculo apoiado no comportamento de espaço de escalas, obtem-se a localização e correção de sub-regiões do edge map que perturbam a deformação. Por outro lado, é incorporada uma nova característica que permite ao algoritmo realizar atividades de classificação. O algoritmo consegue determinar a presença ou ausência de um objeto de interesse utilizando um valor mínimo de deformação. O algoritmo é validado através do tratamento de imagens sintéticas e médicas comparando os resultados com os obtidos no modelo tradicional de contornos deformáveis GVF. / Due to the variation of the quality and noise in medical images, the classic image segmentation techniques are usually ineffective. In this work, we present a new algorithm that is composed of two techniques: the gradient vector flow deformable contours (GVF) and the scale-space technique using a diffusion process. A bibliographical revision of the models that work with deformable contours was accomplished, they were classified in parametric and geometric models. Among the parametric models the gradient vector flow deformable contours (GVF) was chosen. This approach offers precision in the representation of biological structures where other models does not. Thus, the algorithm improves the edge map to guide the deformation using recursive operations. With this estimation based on the behavior of the scale-space techniques it is realized, the localization and correction of sub-areas of the edge map that disturb the deformation. On the other hand, it was incorporated a new characteristic that allows the algorithm to accomplish classification activities. That is, the algorithm determines the presence or absence of a target object using a minimal deformation area. Our method was validated on both, simulated images and medical images making a comparison with the traditional GVF deformable contours.

This dissertation presents the theoretical and experimental exploration of active reflective components specifically for large-aperture adaptive optical zoom systems. An active reflective component can change its focal length by physically deforming its reflecting surface. Adaptive optical zoom (AOZ) utilizes active components in order to change magnification and achieve optical zoom, as opposed to traditional zooming systems that move elements along the optical axis. AOZ systems are theoretically examined using a novel optical design theory that enables a full-scale tradespace analysis, where optical design begins from a broad perspective and optimizes to a particular system. The theory applies existing strategies for telescope design and aberration simulation to AOZ, culminating in the design of a Cassegrain objective with a 3.3X zoom ratio and a 375mm entrance aperture. AOZ systems are experimentally examined with the development of a large-aperture active mirror constructed of a composite material called carbon fiber reinforced polymer (CFRP). The active CFRP mirror uses a novel actuation method to change radius of curvature, where actuators press against two annular rings placed on the mirror's back. This method enables the radius of curvature to increase from 2000mm to 2010mm. Closed-loop control maintains good optical performance of 1.05 waves peak-to-valley (with respect to a HeNe laser) when the active CFRP mirror is used in conjunction with a commercial deformable mirror.

In Chapter 2, we consider the hydrostatic equilibrium figure of a rotating earth model with arbitrary radial density profile. We derive an exact non-linear partial differential equation describing the equilibrium figure. Perturbation theory is used to obtain approximate forms of this equation, and we show that the first-order theory is equivalent to Clairaut's equation. In Chapter 3, a method for parametrizing the possible equilibrium stress fields of a laterally heterogeneous earth model is described. In this method a solution of the equilibrium equations is first found that satisfies some desirable physical property. All other solutions can be written as the sum of this equilibrium stress field and a divergence-free stress tensor field whose boundary tractions vanish. In Chapter 4, we consider the minor vector method for the stable numerical solution of systems of linear ordinary differential equations. Results are presented for the application of the method to the calculation of seismic displacement fields in spherically symmetric, self-gravitating earth models. In Chapter 5, we present a new implementation of the direct solution method for calculating normal mode spectra in laterally heterogeneous earth models. Numerical tests are presented to demonstrate the validity and effectiveness of this method for performing large mode coupling calculations. In Chapter 6, we consider the theoretical basis for the viscoelastic normal mode method which is used in studies of seismic wave propagation, post-glacial rebound, and post-seismic deformation. We show how the time-domain solution to the viscoelastodynamic equation can be written as a normal mode sum in a rigorous manner.

In this thesis, the problem of well-posedness of nonlinear viscoelasticity under the assumptions allowing for phase transformations in solids is considered. In one space dimension we prove existence and uniqueness of the solutions for the quasistatic version of the model using approximating sequences corresponding to the case when initial data takes finitely many values. This special case also provides upper and lower bounds for the solutions which are interesting in their own rights. We also show equivalence of the existence theory we develop with that of gradient flows when the stored-energy function is assumed to be -convex. Asymptotic behaviour of the solutions as time goes to infinity is then investigated and stabilization results are obtained by means of a new argument. Finally, we look at the problem from the viewpoint of curves of maximal slope and follow a time-discretization approach. We introduce a three-dimensional method based on composition of time-increments, as a result of which we are able to deal with the physical requirement of frame-indifference. In order to test this method and distinguish the difficulties for possible generalizations, we look at the problem in a convex setting. At the end we are able to obtain convergence of the minimization scheme as time step goes to zero.

Following trauma, such as an impact injury or stroke, brain tissue can swell. Swelling is the result of water accumulation in the tissue that is driven by pathological changes, such as increased permeability of the capillary walls and osmotic pressure changes within the tissue. Swelling causes increased intracranial pressure and mechanical deformation of the brain tissue, exacerbating the original injury. Furthermore, prolonged local swelling can lead to the spread of damage to the (initially undamaged) surrounding tissue, since compression and increased intracranial pressure may restrict blood flow in this tissue. In this thesis, we develop mathematical models to examine the consequences of pathophysiological damage mechanisms on the swelling, and associated stress and strain, experienced by brain tissue. Mixture theory is used to represent brain tissue as a mixture of elastic solid, fluid and solutes. This modelling approach allows elastic deformations to be coupled with hydrodynamic pressure and osmotic gradients; the consequences of different mechanisms of damage may then be quantified. We consider three particular problems motivated by experimental observations of swelling brain tissue. Firstly, we investigate the swelling of isolated, damaged, brain tissue slices; we show that mechanisms leading to an osmotic pressure difference between the tissue slice and its surroundings can explain experimental observations for swollen tissue slices. Secondly, we use our modelling approach to demonstrate that local changes in capillary permeability can cause significant stresses and strains in the surrounding tissue. Thirdly, we investigate the conditions under which a locally swollen, damaged, region can cause compression of the vasculature within the surrounding tissue, and potentially result in damage propagation. To do this, we propose a coupled model for the oxygen concentration within, and mechanical deformation of, brain tissue. We use our model to assess the impact of treatment strategies on damage propagation through the tissue, and show that performing a craniectomy reduces the extent of propagation.

Biological tissue differs from other materials in many ways. Perhaps the most crucial difference is its ability to grow. Growth processes may give rise to stresses that exist in a body in the absence of applied loads and these are known as residual stresses. Residual stress is present in many biological systems and can have important consequences on the mechanical response of a body. Mathematical models of biological structures must therefore be able to capture accurately the effects of differential growth and residual stress, since greater understanding of the roles of these phenomena may have applications in many fields. In addition to residual stresses, biological structures often have a complex morphology. The theory of 3-D elasticity is analytically tractable in modelling mechanical properties in simple geometries such as a cylinder. On the other hand, rod theory is well-suited for geometrically-complex deformations, but is unable to account for residual stress. In this thesis, we aim to develop a map between the two frameworks. Firstly, we use 3-D elasticity to determine effective mechanical properties of a growing cylinder and map them into an effective rod. Secondly, we consider a growing filament embedded in an elastic foundation. Here, we estimate the degree of transverse reinforcement the foundation confers on the filament in terms of its material properties. Finally, to gain a greater understanding of the role of residual stress in biological structures, we consider a case study: the chameleon's tongue. In particular we consider the role of residual stress and anisotropy in aiding the rapid projection of the tongue during prey capture. We construct a mechanical model of the tongue and use it to investigate a proposed mechanism of projection by means of an energy balance argument.

The application of techniques from nonlinear analysis to materials science has seen great developments in the recent years and it has really been a driving force for substantial mathematical research in the area of partial differential equations and the multi-dimensional calculus of variations. This thesis has been motivated by two recent and remarkable experimental observations of H. Seiner in shape-memory alloys which we attempt to interpret mathematically. Much of the work is original and has given rise to deep problems in the calculus of variations. Firstly, we study the formation of non-classical austenite-martensite interfaces. Ball & Carstensen (1997, 1999) theoretically investigated the possibility of the occurrence of such interfaces and studied the cubic-to-tetragonal case extensively. In this thesis, we present an analysis of non-classical austenite-martensite interfaces recently observed by Seiner et al.~in a single crystal of a CuAlNi shape-memory alloy, undergoing a cubic-to-orthorhombic transition. We show that these can be described by the general nonlinear elasticity model and we make some predictions regarding the admissible volume fractions of the martensitic variants involved, as well as the habit plane normals. Interestingly, in the above experimental observations, the interface between the austenite and the martensitic configuration is never exactly planar, but rather slightly curved, resulting from the pattern of martensite not being exactly homogeneous. However, it is not clear how one can reconstruct the inhomogeneous configuration as a stress-free microstructure and, instead, a theoretical approach is followed. In this approach, a general method is provided for the construction of a compatible curved austenite-martensite interface and, by exploiting the structure of quasiconvex hulls, the existence of curved interfaces is shown in two and three dimensions. As far as the author is aware of, this is the first construction of such a curved austenite-martensite interface. Secondly, we study the nucleation of austenite in a single crystal of a CuAlNi shape-memory alloy consisting of a single variant of stabilized 2H martensite. The nucleation process is induced by localized heating and it is observed that, regardless of where the localized heating is applied, the nucleation points are always located at one of the corners of the sample - a rectangular parallelepiped in the austenite. Using a simplified nonlinear elasticity model, we propose an explanation for the location of the nucleation points by showing that the martensite is a local minimizer of the energy with respect to localized variations in the interior, on faces and edges of the sample, but not at some corners, where a localized microstructure can lower the energy. The result for the interior, faces and edges is established by showing that the free-energy function satisfies a set of quasiconvexity conditions at the stabilized variant throughout the specimen, provided this is suitably cut. The proofs of quasiconvexity are based on a rigidity argument and are specific to the change of symmetry in the phase transformation. To the best of the author's knowledge, quasiconvexity conditions at edges and corners have not been considered before.