This dissertation addresses the problem of adaptive image filtering. Although the topic has a long history in the image processing community, researchers continuously present novel methods to obtain ever better image restoration results. With an expanding market for individuals who wish to share their everyday life on social media, imaging techniques such as compact cameras and smart phones are important factors. Naturally, every producer of imaging equipment desires to exploit cheap camera components while supplying high quality images. One step in this pipeline is to use sophisticated imaging software including, e.g., noise reduction to reduce manufacturing costs, while maintaining image quality. This thesis is based on traditional formulations such as isotropic and tensor-based anisotropic diffusion for image denoising. The difference from main-stream denoising methods is that this thesis explores the effects of introducing contextual information as prior knowledge for image denoising into the filtering schemes. To achieve this, the adaptive filtering theory is formulated from an energy minimization standpoint. The core contributions of this work is the introduction of a novel tensor-based functional which unifies and generalises standard diffusion methods. Additionally, the explicit Euler-Lagrange equation is derived which, if solved, yield the stationary point for the minimization problem. Several aspects of the functional are presented in detail which include, but are not limited to, tensor symmetry constraints and convexity. Also, the classical problem of finding a variational formulation to a given tensor-based partial differential equation is studied. The presented framework is applied in problem formulation that includes non-linear domain transformation, e.g., visualization of medical images. Additionally, the framework is also used to exploit locally estimated probability density functions or the channel representation to drive the filtering process. Furthermore, one of the first truly tensor-based formulations of total variation is presented. The key to the formulation is the gradient energy tensor, which does not require spatial regularization of its tensor components. It is shown empirically in several computer vision applications, such as corner detection and optical flow, that the gradient energy tensor is a viable replacement for the commonly used structure tensor. Moreover, the gradient energy tensor is used in the traditional tensor-based anisotropic diffusion scheme. This approach results in significant improvements in computational speed when the scheme is implemented on a graphical processing unit compared to using the commonly used structure tensor. / VIDI / NACIP / GARNICS / EMC^2

Cette thèse traite des problèmes des concentrations de contraintes locales, en particularité des effets des bords libres dans des structures stratifiés. À l'interface entre deux couches avec des propriétés élastiques différentes, les contraintes ont un comportement singulier dans le voisinage du bord libre en supposant un comportement de matériau élastique linéaire. Par conséquent, ils sont essentiels pour promouvoir le délaminage. Via Formulation unifiée de la Carrera (CUF) différents modèles cinématiques sont testés dans le but de capter les concentrations de contraintes. Dans la première partie de ce travail, les approches de modélisation dimensionnelle réduits sont comparées. Deux classe principale sont présentés: la couche équivalent (ESL) et l'approche par couche, LW. Par la suite leurs capacités à capter les singularités sont comparées. En utilisant une fonction a priori singulière, via une expression exponentielle, une mesure des contraintes singulières est introduite. Seulement deux paramètres décrivent pleinement les composantes des contraintes singulières au voisinage du bord libre. Sur la base des paramètres obtenus les modèles sont comparés et aussi les effets sous des charges d'extension et de flexion et pour différents stratifiés. Les résultats montrent une nécessité des modèles complexes dans le voisinage du bord libre. Cependant loin des bords libres, dans le centre de plaques composite, aucune différence significative ne peut être noté pour les modèles plutôt simples. La deuxième partie de ce travail est donc dédiée au couplage de modèles cinématique incompatibles. Modèles complexes et coûteux sont utilisés seulement dans des domaines locaux d'intérêt, tandis que les modèles économiques simples seront modéliser le domaine global. La eXtended Variational Formulation (XVF) est utilisé pour coupler les modèles de dimensionnalité homogènes mais de cinématique hétérogènes. Ici pas de recouvrement de domaine est présent. En outre, le XVF offre la possibilité d'adapter les conditions imposées à l'interface en utilisant un paramètre scalaire unique. On montre que, pour le problème de dimensionnalité homogène, que deux conditions différentes peuvent être imposées par ce paramètre. Un correspondant à des conditions fortes des Multi Point Constraints (MPC) et un second fournir des conditions faibles. La dernière offre la possibilité de réduire extrêmement le domaine qui utilise le modèle cinématique complexe, sans perte de précision locale. Comme il s'agit de la première application de la XVF vers les structures composites, le besoin d'un nouvel opérateur de couplage a été identifié. Un nouveau formulaire est proposé, testé et sa robustesse sera évaluée. / This work considers local stress concentrations, especially the free-Edge effects of multilayered structures. At the interface of two adjacent layers with different elastic properties, the stresses can become singular in the intermediate vicinity of the free edge. This is valid while assuming a linear elastic material behaviour. As a consequence this zones are an essential delamination trigger. Via the Carrera Unified Formulation (CUF), different kinematical models are testes in order to obtain the correct local stress concentration. In the first part of this work, the reduced dimensional modelling approaches are compared. Two main class are presented: Equivalent Single Layer (ESL) models treating the layered structure like one homogenous plate of equal mechanical proper- ties, and the Layer Wise approach, treating each layer independently. Subsequently their capabilities to capture the appearing singularities are compared. In order to have a comparable measurement of those singularities, the obtained stress distributions will be expressed via a power law function, which has a priori a singular behaviour. Only two parameters fully describe therefore the singular stress components in the vicinity of the free edge. With the help of these two parameters not only the different models capabilities will be compared, but also the free edge effect itself will be measured and compared for different symmetrical laminates and the case of extensional and uniform bending load. The results for all laminates under both load cases confirm the before stated need for rather complex models in the vicinity of the free edge. However far from the free edges, in the composite plates centre, no significant difference can be noted for rather simple models. The second part of this work is therefore dedicated to the coupling of kinematically incompatible models. The use of costly expensive complex models is restricted to local domains of interest, while economic simple models will model the global do- main. The Extended Variational Formulation (XVF) is identified as the most suitable way to couple the kinematically heterogenous but dimensional homogenous models. As it uses a configuration with one common interface without domain overlap, the additional efforts for establishing the coupling are limited. Further the XVF offers the possibility to adapt the conditions imposed at the interface using a single scalar parameter. It will be shown that for the homogenous dimensional problem under consideration only two different conditions can be imposed by this parameter. One matching the strong conditions imposed by the classical Multi Point Constrains (MPC) and a second one providing a weak condition. The last one is shown to provide the possibility to reduce further the domain using the complex kinematical model, without the loss of local precision. As this is the first application of the XVF towards composite structures, the need for a new coupling operator was identified. A new form is proposed, tested and its robustness will be evaluated.

Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro / Este trabalho apresenta uma modelagem matemática para o processo de aquecimento de um corpo exposto a uma fonte pontual de radiação térmica. O resultado original que permite a solução exata de uma equação diferencial parcial não linear a partir de uma seqüência de problemas lineares também é apresentado. Gráficos gerados com resultados obtidos pelo método de diferenças finitas ilustram a solução do problema proposto. / This work presents a mathematical model for the heating process on a body exposed to a punctual source of thermal radiation. An original result, that allows the construction of the exact solution for a non-linear partial differential equation by solving a sequence of linear problems, is also presented. Graphic images generated from the results obtained through the Finite Difference Method illustrate the solution of the proposed problem.

Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro / Este trabalho apresenta uma modelagem matemática para o processo de aquecimento de um corpo exposto a uma fonte pontual de radiação térmica. O resultado original que permite a solução exata de uma equação diferencial parcial não linear a partir de uma seqüência de problemas lineares também é apresentado. Gráficos gerados com resultados obtidos pelo método de diferenças finitas ilustram a solução do problema proposto. / This work presents a mathematical model for the heating process on a body exposed to a punctual source of thermal radiation. An original result, that allows the construction of the exact solution for a non-linear partial differential equation by solving a sequence of linear problems, is also presented. Graphic images generated from the results obtained through the Finite Difference Method illustrate the solution of the proposed problem.

This doctorial thesis deals with the derivation of a mechanical model for the simulation of the load-bearing behavior of a strengthening layer made of textile reinforced concrete to strengthen a shell structure. The main focus lies on both the geometrical and physical non-linear three-dimensional shell formulation and on its transfer into an efficient finite element. The distinctive feature of the presented shell formulation is its surface-orientation. This enables the analysis of a strengthening layer applied on one of the outer surfaces of a structure in a very natural way, since a problem-oriented mechanical modeling is achieved. Hereby, in contrast to classical shell theories the three-dimensionality of the material behavior's description can completely maintained. This is necessary, since a more accurate modeling of the material behavior of textile reinforced concrete can only be done three-dimensional. Within the scope of this thesis an anisotropic hyper elastic constitutive relation is given in order to obtain a first approximation of textile reinforced concrete's material behavior against the background of material theory. Furthermore a specification of the hyper elastic constitutive relation is obtained upon the basis of the principle of material symmetry and is prepared for the application in the shell formulation. The numerical solution of the field problem necessitates the transfer of the surface-related shell formulation into a two-dimensional variational formulation in order to obtain a sound mathematical starting point for the conversion into an efficient volume shell element. For the realization of a numerical efficient finite element an element formulation using a low-order ansatz should be favored. Because of reducing the number of degrees of freedom and therewith of possible deformation modes, artificial stiffening effects will appear. A way out is given by a special extention of the assumed natural strain and the assumed strain method for the utilization in the presented surface-related shell formulation. This leads to a slightly higher numerical effort, but allows a reliable and efficient finite element formulation finally verified in distinct meaningful non-linear simulations. / Die Herleitung eines mechanischen Modells zur numerischen Simulation des Tragverhaltens einer Verstärkungsschicht für Flächentragwerke aus textilbewehrtem Feinbeton ist Gegenstand dieser Arbeit. Hierbei liegt das Hauptaugenmerk auf einer sowohl geometrisch als auch physikalisch nichtlinearen dreidimensionalen Schalenformulierung und deren Umsetzung in ein effizientes finites Element. Die Besonderheit der hier vorgestellten Schalenformulierung ist deren Oberflächenbezug, der es auf natürliche Weise ermöglicht, eine auf die Struktur aufgebrachte Verstärkungsschicht zu berechnen. Diese Verfahrensweise ist als eine der Problemstellung angepaßte mechanische Modellbildung anzusehen. Hierbei kann im Gegensatz zu klassischen Schalentheorien die Dreidimensionalität der Materialbeschreibung vollständig aufrecht erhalten und damit die verfeinerte Erfassung des Materialverhaltens von Textilbeton ermöglicht werden, die nur dreidimensional erfolgen kann. Eine materialtheoretische Herleitung von anisotropen hyper-elastischen konstitutiven Beziehungen als erste Approximation zur Erfassung des Materialverhaltens von textilbewehrtem Feinbeton wird vorgestellt und deren Spezifizierung auf der Basis des Prinzips der materiellen Symmetrie durchgeführt sowie zur Anwendung in der Schalenformulierung aufbereitet. Die numerische Lösung des Feldproblems erfordert die Umsetzung der oberflächenorientierten Schalenformulierung in eine zweidimensionale Variationsformulierung mit dem Ziel, einen mathematisch fundierten Ausgangspunkt für die Entwicklung eines leistungsfähigen Volumen"=Schalen"=Elementes zu erhalten. Zur Realisierung eines numerisch effizienten finiten Elementes ist eine Elementformulierung mit möglichst geringer Ansatzordnung verwendet worden. Da hierdurch die Freiheitsgrade des Elementes und damit die möglichen Verformungsmodi in ihrer Anzahl eingeschränkt werden, sind künstliche Versteifungseffekte zu erwarten. Einen Ausweg bieten hier spezielle Erweiterungen der Assumed Natural Strain und der Enhanced Assumed Strain Methode für die hier vorliegende oberflächenorientierte Schalenformulierung. Dies erhöht den numerischen Aufwand unwesentlich, ermöglicht jedoch insgesamt eine zuverlässige und effiziente Elementformulierung, deren Brauchbarkeit abschließend in verschiedenen aussagekräftigen nichtlinearen Simulationen nachgewiesen wird.

Este trabalho estuda a transferência de calor por condução considerando a condutividade térmica como uma função constante por partes da temperatura. Esta relação, embora fisicamente mais realista que supor a condutividade térmica constante, permite obter uma forma explícita bem simples para a inversa da Transformada de Kirchhoff (empregada para tratar a não linearidade do problema). Como exemplo, apresenta-se uma solução exata para um problema com simetria esférica. Em seguida, propôe-se uma formulação variacional (com unicidade demonstrada) que introduz um funcional cuja minimização é equivalente à solução do problema na forma forte. Finalmente compara-se uma solução exata obtida pela inversa da Transformada de Kirchhoff com a solução obtida via formulação variacional. / This work studies conduction heat transfer considering thermal conductivity as a piecewise constant function of temperature. This relationship, although physically more realistic than assuming constant thermal conductivity, provides a simple explicit form for the inverse of Kirchhoff transformation (employed to deal with the problem non-linearity). An exact solution for a problem with spherical symmetry is presented, as an example. In the sequence, a variational formulation (with demonstrated uniqueness) is proposed. This formulation introduces a functional whose minimization is equivalent to the solution of the problem in the strong form. Finally an exact solution obtained using the inverse of Kirchhoff transformation is compared with the solution obtained via variational formulation.

Este trabalho estuda a transferência de calor por condução considerando a condutividade térmica como uma função constante por partes da temperatura. Esta relação, embora fisicamente mais realista que supor a condutividade térmica constante, permite obter uma forma explícita bem simples para a inversa da Transformada de Kirchhoff (empregada para tratar a não linearidade do problema). Como exemplo, apresenta-se uma solução exata para um problema com simetria esférica. Em seguida, propôe-se uma formulação variacional (com unicidade demonstrada) que introduz um funcional cuja minimização é equivalente à solução do problema na forma forte. Finalmente compara-se uma solução exata obtida pela inversa da Transformada de Kirchhoff com a solução obtida via formulação variacional. / This work studies conduction heat transfer considering thermal conductivity as a piecewise constant function of temperature. This relationship, although physically more realistic than assuming constant thermal conductivity, provides a simple explicit form for the inverse of Kirchhoff transformation (employed to deal with the problem non-linearity). An exact solution for a problem with spherical symmetry is presented, as an example. In the sequence, a variational formulation (with demonstrated uniqueness) is proposed. This formulation introduces a functional whose minimization is equivalent to the solution of the problem in the strong form. Finally an exact solution obtained using the inverse of Kirchhoff transformation is compared with the solution obtained via variational formulation.

In recent years there have been a major developments in the computational mechanics community when it comes to modelling of fracture mechanics. There are now several well established numerical methods that are implemented in commercial programs such as: Phase Field Modelling, Enhanced Assumed Strain (EAS), Smeared Crack Methods, Extended Finite Element Method (XFEM) and Partition of Unity Finite Element Method (PUFEM). This work has been based on PUFEM and it has been of interest to investigate if this numerical method can be combined with plastic deformation. The reason for this is that it has been known that complex structural and material phenomenon such as embrittlement of steels or composites show a variation of fracture toughness. Thus more advanced approaches are required to capture the response of such structures and materials. The analysis was split in to three parts. First, a benchmark analysis was done with linear tetrahedral elements and, which has been implemented in previous applications. Thereafter the same analysis was done for quadratic tetrahedral elements. Finally, the interface between plasticity and PUFEM was implemented and was analysed with the same geometries as in the benchmark case. The analysis show that it is possible to combine plasticity in PUFEM setting and that it has a possibility to be used in future applications. / Under dem senaste åren har det skett en massiv utveckling inom beräkningsmekaniken när det kommer till att modellera brottmekaniska fenomen. Det finns nu ett flertal väletablerade numeriska metoder som är implementerade i kommersiella program såsom: Phase Field Modelling, Enhanced Assumed Strain (EAS), Smeared Crack Methods, Extended Finite Element Method (XFEM) and Partition of Unity Finite Element Method (PUFEM). Detta arbete har fokuserat på (PUFEM) och det har varit av intresse att se om denna metod kan kombineras tillsammans med plastisk deformation. Anledningen till att detta har varit av intresse är på grund av att det finns ett flertal komplexa struktur - och materialfenomen såsom försprödning av stål och kompositer som uppvisar en variation i sin duktilitet. Detta medför att mer avancerade metoder behöver tillämpas för att fånga responsen av dessa strukturer och material. Analysen som har genomförts var uppdelad i tre delar. Först analyserades ett problem med linjär tetrahedriska element som ett riktmärke, detta har implementerats i tidigare applikationer. Därefter analyserades samma problem fast med tethraderiska element med kvadratisk interpolation. Slutligen så integrerades plasticitet med PUFEM, där samma geometrier analyserades som riktmärke. Den analysen som har genomfört visar att det går att kombinera plastiska deformation tillsammans med PUFEM och att denna metod har potentialen att användas i framtida applikationer.

Este trabalho trata da combinação entre a formulação híbrida-mista de tensão (FHMT) (Freitas et al. (1996)), para a elasticidade plana, com o método dos elementos finitos generalizados (MEFG), Duarte et al. (2000). O MEFG se caracteriza como uma forma não-convencional do método dos elementos finitos (MEF) que resulta da incorporação a este de conceitos e técnicas dos métodos sem malha, como o enriquecimento nodal proposto do método das nuvens “hp". Como na FHMT são aproximados dois campos no domínio (tensão e deslocamento) e um no contorno (deslocamento), diferentes possibilidades de enriquecimento nodal são exploradas. Para a discretização do modelo híbrido-misto empregam-se elementos finitos quadrilaterais com funções de forma bilineares para o domínio e elementos lineares para o contorno. Essas funções são enriquecidas por funções polinomiais, trigonométricas, polinômios que proporcionam distribuição de tensões auto-equilibradas ou mesmo funções especiais relacionadas às soluções dos problemas de fratura. Uma extensão do teste numérico abordado em Zienkiewicz et al. (1986) é proposta como investigação inicial das condições necessárias para garantia de estabilidade da resposta numérica. O estudo da estabilidade é completado com a análise da condição de Babuška-Brezzi (inf-sup). Esta condição é aplicada nos elementos finitos quadrilaterais híbridos-mistos enriquecidos por meio de um teste numérico, denominado de inf-sup teste, desenvolvido com base no trabalho de Chapelle e Bathe (1993). Exemplos numéricos revelam que a FHMT é uma interessante alternativa para obtenção de boas estimativas para os campos de tensões e deslocamentos, usando-se enriquecimento sobre alguns nós de malhas pouco refinadas / This work presents a combination of hybrid-mixed stress model formulation (HMSMF) (Freitas et al. (1996)), to treat plane elasticity problems, with generalized finite element method (GFEM), (Duarte et al. (2000)). GFEM is characterized as a nonconventional formulation of the finite element method (FEM). GFEM is the result of the incorporation of concepts and techniques from meshless methods. One example of these techniques is the nodal enrichment that was formulated in the “hp" clouds method. Since two fields in domain (stress and displacement) and one in boundary (displacement) are approximated in the HMSMF, different possibilities of nodal enrichment are tested. For the discretization of the hybrid-mixed model quadrilateral finite elements with bilinear shape functions for the domain and linear elements for the boundary were employed. These functions are enriched with polynomial functions, trigonometric functions, polynomials that generate self-equilibrated stress distribution, or, even special functions connected with solutions of fracture problems. An extension of the numerical test cited in Zienkiewicz et al. (1986) is proposed as initial investigation of necessary conditions to assure the stability of the numerical answer. The stability study is completed with the analysis of the Babuška-Brezzi (inf-sup) condition. This last condition is applied to hybrid-mixed enrichment quadrilaterals finite elements by means of a numerical test, denominated inf-sup test, which was developed based on paper of Chapelle and Bathe (1993). Numerical examples reveal that HMSMF is an interesting alternative to obtain good estimates of the stress and displacement fields, using enrichment over some nodes of poor meshes

Este trabalho trata da combinação entre a formulação híbrida-mista de tensão (FHMT) (Freitas et al. (1996)), para a elasticidade plana, com o método dos elementos finitos generalizados (MEFG), Duarte et al. (2000). O MEFG se caracteriza como uma forma não-convencional do método dos elementos finitos (MEF) que resulta da incorporação a este de conceitos e técnicas dos métodos sem malha, como o enriquecimento nodal proposto do método das nuvens “hp”. Como na FHMT são aproximados dois campos no domínio (tensão e deslocamento) e um no contorno (deslocamento), diferentes possibilidades de enriquecimento nodal são exploradas. Para a discretização do modelo híbrido-misto empregam-se elementos finitos quadrilaterais com funções de forma bilineares para o domínio e elementos lineares para o contorno. Essas funções são enriquecidas por funções polinomiais, trigonométricas, polinômios que proporcionam distribuição de tensões auto-equilibradas ou mesmo funções especiais relacionadas às soluções dos problemas de fratura. Uma extensão do teste numérico abordado em Zienkiewicz et al. (1986) é proposta como investigação inicial das condições necessárias para garantia de estabilidade da resposta numérica. O estudo da estabilidade é completado com a análise da condição de Babuška-Brezzi (inf-sup). Esta condição é aplicada nos elementos finitos quadrilaterais híbridos-mistos enriquecidos por meio de um teste numérico, denominado de inf-sup teste, desenvolvido com base no trabalho de Chapelle e Bathe (1993). Exemplos numéricos revelam que a FHMT é uma interessante alternativa para obtenção de boas estimativas para os campos de tensões e deslocamentos, usando-se enriquecimento sobre alguns nós de malhas pouco refinadas / This work presents a combination of hybrid-mixed stress model formulation (HMSMF) (Freitas et al. (1996)), to treat plane elasticity problems, with generalized finite element method (GFEM), (Duarte et al. (2000)). GFEM is characterized as a nonconventional formulation of the finite element method (FEM). GFEM is the result of the incorporation of concepts and techniques from meshless methods. One example of these techniques is the nodal enrichment that was formulated in the “hp” clouds method. Since two fields in domain (stress and displacement) and one in boundary (displacement) are approximated in the HMSMF, different possibilities of nodal enrichment are tested. For the discretization of the hybrid-mixed model quadrilateral finite elements with bilinear shape functions for the domain and linear elements for the boundary were employed. These functions are enriched with polynomial functions, trigonometric functions, polynomials that generate self-equilibrated stress distribution, or, even special functions connected with solutions of fracture problems. An extension of the numerical test cited in Zienkiewicz et al. (1986) is proposed as initial investigation of necessary conditions to assure the stability of the numerical answer. The stability study is completed with the analysis of the Babuška-Brezzi (inf-sup) condition. This last condition is applied to hybrid-mixed enrichment quadrilaterals finite elements by means of a numerical test, denominated inf-sup test, which was developed based on paper of Chapelle and Bathe (1993). Numerical examples reveal that HMSMF is an interesting alternative to obtain good estimates of the stress and displacement fields, using enrichment over some nodes of poor meshes