Research in multi-touch interaction has typically been focused on direct spatial manipulation; techniques have been created to result in the most intuitive mapping between the movement of the hand and the resultant change in the virtual object. As we attempt to design for more complex operations, the effectiveness of spatial manipulation as a metaphor becomes weak. We introduce two new platforms for multi-touch computing: a gesture recognition system, and a new interaction technique. I present Multi-Tap Sliders, a new interaction technique for operation in what we call non-spatial parametric spaces. Such spaces do not have an obvious literal spatial representation, (Eg.: exposure, brightness, contrast and saturation for image editing). The multi-tap sliders encourage the user to keep her visual focus on the tar- get, instead of requiring her to look back at the interface. My research emphasizes ergonomics, clear visual design, and fluid transition between modes of operation. Through a series of iterations, I develop a new technique for quickly selecting and adjusting multiple numerical parameters. Evaluations of multi-tap sliders show improvements over traditional sliders. To facilitate further research on multi-touch gestural interaction, I developed mGestr: a training and recognition system using hidden Markov models for designing a multi-touch gesture set. Our evaluation shows successful recognition rates of up to 95%. The recognition framework is packaged into a service for easy integration with existing applications.

Subsurface flow phenomena characterize many important societal issues in energy and the environment. A key feature of these problems is that subsurface properties are uncertain, due to the sparsity of direct observations of the subsurface. The Bayesian formulation of this inverse problem provides a systematic framework for inferring uncertainty in the properties given uncertainties in the data, the forward model, and prior knowledge of the properties. We address the problem: given noisy measurements of the head, the pdf describing the noise, prior information in the form of a pdf of the hydraulic conductivity, and a groundwater flow model relating the head to the hydraulic conductivity, find the posterior probability density function (pdf) of the parameters describing the hydraulic conductivity field. Unfortunately, conventional sampling of this pdf to compute statistical moments is intractable for problems governed by large-scale forward models and high-dimensional parameter spaces. We construct a Gaussian process surrogate of the posterior pdf based on Bayesian interpolation between a set of "training" points. We employ a greedy algorithm to find the training points by solving a sequence of optimization problems where each new training point is placed at the maximizer of the error in the approximation. Scalable Newton optimization methods solve this "optimal" training point problem. We tailor the Gaussian process surrogate to the curvature of the underlying posterior pdf according to the Hessian of the log posterior at a subset of training points, made computationally tractable by a low-rank approximation of the data misfit Hessian. A Gaussian mixture approximation of the posterior is extracted from the Gaussian process surrogate, and used as a proposal in a Markov chain Monte Carlo method for sampling both the surrogate as well as the true posterior. The Gaussian process surrogate is used as a first stage approximation in a two-stage delayed acceptance MCMC method. We provide evidence for the viability of the low-rank approximation of the Hessian through numerical experiments on a large scale atmospheric contaminant transport problem and analysis of an infinite dimensional model problem. We provide similar results for our groundwater problem. We then present results from the proposed MCMC algorithms. / text

No description available.

Visão é um processo que envolve uma grande quantidade de informações, as quais precisam ser otimizadas de alguma forma para propiciar um processamento eficiente. Grande parte das informações visuais estão contidas nos contornos de uma imagem e uma grande redução no volume dos dados pode ser conseguida com a análise dos contornos. Além dos contornos, a detecção de segmentos de reta é o próximo passo na compressão das informações visuais. A detecção de retas ocorre no sistema visual humano, e também no de outros seres vivos. Entre os invertebrados terrestres, o melhor sistema de visão é o das aranhas da família Salticidae e este apresenta características que facilitam a detecção de retas. Este trabalho propõe um novo método de detecção de retas, baseado no sistema visual das aranhas saltadoras, que aborda este problema através de um enfoque inédito, por otimização. O método realiza a busca por retas através de janelas lineares. Para isso, a detecção de retas será feita em um espaço de parâmetros, com a utilização do algoritmo de maximização de funções \"Downhill Simplex\". O método desenvolvido leva em consideração a natureza discreta da imagem e do espaço de parâmetros utilizado, e este trabalho inclui um estudo detalhado destes espaços discretos. O método incorpora, para lidar adequadamente com as peculiaridades do problema, características como \"Simulated Annealing\" e largura adaptativa. O desempenho do método depende de um conjunto de parâmetros cujo comportamento é de difícil previsão, e a escolha de um conjunto foi realizada através de um algoritmo genético. O trabalho envolve também a construção de um protótipo para a realização de testes utilizando o método desenvolvido. Os resultados foram analisados quanto a precisão na detecção de retas, ao tempo de processamento e a movimentação das janelas lineares, relacionada aos esforços na busca por retas. / Vision is a process that involves a large amount of information that need to be somehow optimized to allow efficient processing. Most of the visual information is contained in the contours of an image and a considerable reduction in the amount of data can be achieved by fmding and processing these contours. The next step to further compress the visual data is to fmd straight segments, and represent the contours in terms of these entities. Straight-line segment detection is performed by the human visual system, as well as by other creatures. Among terrestrial invertebrates, the best visual system is that of the Salticidae family of spiders, also known as jumping spiders. This visual system presents some characteristics that facilitate the detection of straight-lines. The present work proposes a new method for straight-line detection, based on the visual system of the jumping spiders, using linear windows. This method approach the straight-line detection problem through an optimization point of view yet unexplored in literature. The detection will be accomplished in a parameter space, using the \"Downhill Simplex\" maximization algorithm. The method considers the discrete nature of both the image and the parameter spaces, and this work includes a detailed analysis of these discrete spaces. The method also incorporates, to adequately deal with the specific characteristics of the problem, resources such as \"Simulated Annealing\" and adaptive width of the linear windows. The performance of the method depends on a set of parameters, which behavior is hard to predict, and the choice of an adequate set was made using a genetic algorithm. The work also involves the project and construction of a prototype, to evaluate the proposed method. Results were analyzed regarding their precision, processing time and the movements of the linear windows, related to the effort made to detect the straight lines.

Within this work an approach was developed, which utilises airborne laser scanner data in order to generate 3D building models. These 3D building models may be used for technical and environmental planning. The approach has to follow certain requirements such as working automatically and robust and being flexible in use but still practicable. The approach starts with small point clouds containing one building at the time extracted from laser scanner data set by applying a pre-segmentation scheme. The laser scanner point cloud of each building is analysed separately. A 2.5D-Delaunay triangle mesh structure (TIN) is calculated into the laser scanner point cloud. For each triangle the orientation parameters in space (orientation, slope and perpendicular distance to the barycentre of the laser scanner point cloud) are determined and mapped into a parameter space. As buildings are composed of planar features, primitives, triangles representing these features should group in parameter space. A cluster analysis technique is utilised to find and outline these groups/clusters. The clusters found in parameter space represent plane objects in object space. Grouping adjacent triangles in object space - which represent points in parameter space - enables the interpolation of planes in the ALS points that form the triangles. In each cluster point group a plane in object space is interpolated. All planes derived from the data set are intersected with their appropriate neighbours. From this, a roof topology is established, which describes the shape of the roof. This ensures that each plane has knowledge on its direct adjacent neighbours. Walls are added to the intersected roof planes and the virtual 3D building model is presented in a file written in VRML (Virtual Reality Macro Language). Besides developing the 3D building model reconstruction scheme, this research focuses on the geometric reconstruction and the derivation of attributes of 3D building models. The developed method was tested on different data sets obtained from different laser scanner systems. This study will also show, which potential and limits the developed method has when applied to these different data sets. / In der vorliegenden Arbeit wird eine neue Methode zur automatischen Rekonstruktion von 3D Gebäudemodellen aus Flugzeuglaserscannerdaten vorgestellt. Diese 3D Gebäudemodelle können in technischer und landschaftsplanerischer Hinsicht genutzt werden. Bezüglich der zu entwickelnden Methode wurden Regelungen und Bedingungen erstellt, die eine voll automatische und robuste Arbeitsweise sowie eine flexible und praktikable Nutzung gewährleisten sollten. Die entwickelte Methode verwendet Punktwolken, welche mittels einer Vorsegmentierung aus dem gesamten Laserscannerdatensatz extrahiert wurden und jeweils nur ein Gebäude beinhalten. Diese Laserscannerdatenpunktwolken werden separat analysiert. Eine 2,5D-Delaunay-Dreiecksvermaschung (TIN) wird in jede Punktwolke gerechnet. Für jedes Dreieck dieser Vermaschung werden die Lageparameter im Raum (Ausrichtung, Neigungsgrad und senkrechter Abstand der Ebene des Dreiecks zum Schwerpunkt der Punktwolke) bestimmt und in einen Parameterraum aufgetragen. Im Parameterraum bilden diejenigen Dreiecke Gruppen, welche sich im Objektraum auf ebenen Flächen befinden. Mit der Annahme, dass sich ein Gebäude aus ebenen Flächen zusammensetzt, dient die Identifizierung von Clustern im Parameterraum der Detektierung dieser Flächen. Um diese Gruppen/Cluster aufzufinden wurde eine Clusteranalysetechnik genutzt. Über die detektierten Cluster können jene Laserscannerpunkte im Objektraum bestimmt werden, die eine Dachfläche formen. In die Laserscannerpunkte der somit gefundenen Dachflächen werden Ebenen interpoliert. Alle abgeleiteten Ebenen gehen in den entwickelten Rekonstruktionsalgorithmus ein, der eine Topologie zwischen den einzelnen Ebenen aufbaut. Anhand dieser Topologie erhalten die Ebenen ?Kenntnis? über ihre jeweiligen Nachbarn und können miteinander verschnitten werden. Der fertigen Dachgestalt werden Wände zugefügt und das komplette 3D Gebäudemodell wird mittels VRML (Virtual Reality Macro Language) visualisiert. Diese Studie bezieht sich neben der Entwicklung eines Schemas zu automatischen Gebäuderekonstruktion auch auf die Ableitung von Attributen der 3D Gebäudemodellen. Die entwickelte Methode wurde an verschiedenen Flugzeuglaserscannerdatensätzen getestet. Es wird gezeigt, welche Potentiale und Grenzen die entwickelte Methode bei der Bearbeitung dieser verschiedenen Laserscannerdatensätze hat.

Made available in DSpace on 2016-12-12T20:15:54Z (GMT). No. of bitstreams: 1 pre_textuais.pdf: 54354 bytes, checksum: 81cdd6a9b6a26b1b93efdeee869b2de9 (MD5) Previous issue date: 2011-07-26 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / In this work we study the self-organization of periodic structures on parameter-spaces of the largest Lyapunov exponent (Lyapunov diagrams) of the Watt governor system model. A hierarchical organization and period-adding bifurcation cascades of the periodic structures are observed, and these self-organized cascades accumulate on a periodic boundary. We also show that the periods of the structures organize themselves obeying the solutions of a diophantine equation. / Neste trabalho estudamos a auto-organização de estruturas periódicas no espaço de parâmetros do maior expoente de Lyapunov (diagramas de Lyapunov) em um modelo do sistema regulador deWatt. Uma organização hierárquica e cascatas de bifurcação por adição de período das estruturas periódicas são observadas e estas cascatas auto-organizadas se acumulam em fronteiras periódicas. Também mostramos que os períodos das estruturas organizam-se obedecendo as soluções de equações diofantina

Made available in DSpace on 2016-12-12T20:15:49Z (GMT). No. of bitstreams: 1 Juliana Vicente.pdf: 26659692 bytes, checksum: 42c72bc287e017fe62aec1549942e15b (MD5) Previous issue date: 2013-03-01 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The Fitzhugh-Nagumo model originally consists of two non-linear differential equations, which simulate the behavior of nerve impulse conduction through the neuronal membrane. In this work, we study the dynamical behavior of coupled neuron networks modeled by equations of Fitzhugh-Nagumo, in the numerical viewpoint. The numerical simulations were made considering networks of two, three and four neurons coupled unidirectionally and bidirectionally, for which parameter spaces, isoperiodic diagrams and bifurcation diagrams were built. In the parameter spaces and isoperiodic diagrams we investigated the dynamics of the variation between coupling strength of the neural systems and any other parameters of system, calculating the Lyapunov exponents and the local maximum of a variable respectively. The results showed the existence of periodic structures, self-organized arranged sequentially in a period-adding bifurcation cascade immersed in chaotic regions. Internally these structures undergo bifurcations by period-doubling. In the case of the two neurons model unidirectionally coupled, we observe the formation of periodic structures arranged in period-adding bifurcation cascade. Finally we show that in some cases the networks of three and four neurons the occurrence of hyperchaos, were observed in the dynamics of the systems. / O modelo de Fitzhugh-Nagumo é composto originalmente por um sistema de duas equações diferenciais não-lineares, que simulam o comportamento de condução do impulso nervoso através da membrana neural. Neste trabalho estudamos numericamente o comportamento dinâmico de redes de neurônios acoplados, modeladas pelas equações de Fitzhugh-Nagumo. Consideramos redes de dois, três e quatro neurônios acoplados unidirecionalmente e bidirecionalmente, para as quais foram construídos espaços de parâmetros dos expoentes de Lyapunov, diagramas isoperíodos e diagramas de bifurcação. Nos espaços de parâmetros e diagramas isoperiódicos investigamos a dinâmica da variação entre a intensidade de acoplamento dos sistemas de neurônios e um outro parâmetros quaisquer do sistema, calculando o espectro de Lyapunov e os máximos locais de uma variável, respectivamente. Os resultados evidenciaram a existência de estruturas auto-similares arranjadas sequencialmente em cascatas de bifurcação por adição de período imersas em regiões caóticas. Internamente, as estruturas periódicas exibem cascatas de bifurcações por dobramento de período. No caso do modelo para dois neurônios acoplados unidirecionalmente existe a formação de estruturas periódicas em camada, arranjadas em cascatas de bifurcação por adição de período. Finalmente mostramos que, para as redes de três e quatro neurônios, existe a ocorrência de regiões de hipercaos na dinâmica dos sistemas.

Visão é um processo que envolve uma grande quantidade de informações, as quais precisam ser otimizadas de alguma forma para propiciar um processamento eficiente. Grande parte das informações visuais estão contidas nos contornos de uma imagem e uma grande redução no volume dos dados pode ser conseguida com a análise dos contornos. Além dos contornos, a detecção de segmentos de reta é o próximo passo na compressão das informações visuais. A detecção de retas ocorre no sistema visual humano, e também no de outros seres vivos. Entre os invertebrados terrestres, o melhor sistema de visão é o das aranhas da família Salticidae e este apresenta características que facilitam a detecção de retas. Este trabalho propõe um novo método de detecção de retas, baseado no sistema visual das aranhas saltadoras, que aborda este problema através de um enfoque inédito, por otimização. O método realiza a busca por retas através de janelas lineares. Para isso, a detecção de retas será feita em um espaço de parâmetros, com a utilização do algoritmo de maximização de funções \"Downhill Simplex\". O método desenvolvido leva em consideração a natureza discreta da imagem e do espaço de parâmetros utilizado, e este trabalho inclui um estudo detalhado destes espaços discretos. O método incorpora, para lidar adequadamente com as peculiaridades do problema, características como \"Simulated Annealing\" e largura adaptativa. O desempenho do método depende de um conjunto de parâmetros cujo comportamento é de difícil previsão, e a escolha de um conjunto foi realizada através de um algoritmo genético. O trabalho envolve também a construção de um protótipo para a realização de testes utilizando o método desenvolvido. Os resultados foram analisados quanto a precisão na detecção de retas, ao tempo de processamento e a movimentação das janelas lineares, relacionada aos esforços na busca por retas. / Vision is a process that involves a large amount of information that need to be somehow optimized to allow efficient processing. Most of the visual information is contained in the contours of an image and a considerable reduction in the amount of data can be achieved by fmding and processing these contours. The next step to further compress the visual data is to fmd straight segments, and represent the contours in terms of these entities. Straight-line segment detection is performed by the human visual system, as well as by other creatures. Among terrestrial invertebrates, the best visual system is that of the Salticidae family of spiders, also known as jumping spiders. This visual system presents some characteristics that facilitate the detection of straight-lines. The present work proposes a new method for straight-line detection, based on the visual system of the jumping spiders, using linear windows. This method approach the straight-line detection problem through an optimization point of view yet unexplored in literature. The detection will be accomplished in a parameter space, using the \"Downhill Simplex\" maximization algorithm. The method considers the discrete nature of both the image and the parameter spaces, and this work includes a detailed analysis of these discrete spaces. The method also incorporates, to adequately deal with the specific characteristics of the problem, resources such as \"Simulated Annealing\" and adaptive width of the linear windows. The performance of the method depends on a set of parameters, which behavior is hard to predict, and the choice of an adequate set was made using a genetic algorithm. The work also involves the project and construction of a prototype, to evaluate the proposed method. Results were analyzed regarding their precision, processing time and the movements of the linear windows, related to the effort made to detect the straight lines.

Within this work an approach was developed, which utilises airborne laser scanner data in order to generate 3D building models. These 3D building models may be used for technical and environmental planning. The approach has to follow certain requirements such as working automatically and robust and being flexible in use but still practicable. The approach starts with small point clouds containing one building at the time extracted from laser scanner data set by applying a pre-segmentation scheme. The laser scanner point cloud of each building is analysed separately. A 2.5D-Delaunay triangle mesh structure (TIN) is calculated into the laser scanner point cloud. For each triangle the orientation parameters in space (orientation, slope and perpendicular distance to the barycentre of the laser scanner point cloud) are determined and mapped into a parameter space. As buildings are composed of planar features, primitives, triangles representing these features should group in parameter space. A cluster analysis technique is utilised to find and outline these groups/clusters. The clusters found in parameter space represent plane objects in object space. Grouping adjacent triangles in object space - which represent points in parameter space - enables the interpolation of planes in the ALS points that form the triangles. In each cluster point group a plane in object space is interpolated. All planes derived from the data set are intersected with their appropriate neighbours. From this, a roof topology is established, which describes the shape of the roof. This ensures that each plane has knowledge on its direct adjacent neighbours. Walls are added to the intersected roof planes and the virtual 3D building model is presented in a file written in VRML (Virtual Reality Macro Language). Besides developing the 3D building model reconstruction scheme, this research focuses on the geometric reconstruction and the derivation of attributes of 3D building models. The developed method was tested on different data sets obtained from different laser scanner systems. This study will also show, which potential and limits the developed method has when applied to these different data sets. / In der vorliegenden Arbeit wird eine neue Methode zur automatischen Rekonstruktion von 3D Gebäudemodellen aus Flugzeuglaserscannerdaten vorgestellt. Diese 3D Gebäudemodelle können in technischer und landschaftsplanerischer Hinsicht genutzt werden. Bezüglich der zu entwickelnden Methode wurden Regelungen und Bedingungen erstellt, die eine voll automatische und robuste Arbeitsweise sowie eine flexible und praktikable Nutzung gewährleisten sollten. Die entwickelte Methode verwendet Punktwolken, welche mittels einer Vorsegmentierung aus dem gesamten Laserscannerdatensatz extrahiert wurden und jeweils nur ein Gebäude beinhalten. Diese Laserscannerdatenpunktwolken werden separat analysiert. Eine 2,5D-Delaunay-Dreiecksvermaschung (TIN) wird in jede Punktwolke gerechnet. Für jedes Dreieck dieser Vermaschung werden die Lageparameter im Raum (Ausrichtung, Neigungsgrad und senkrechter Abstand der Ebene des Dreiecks zum Schwerpunkt der Punktwolke) bestimmt und in einen Parameterraum aufgetragen. Im Parameterraum bilden diejenigen Dreiecke Gruppen, welche sich im Objektraum auf ebenen Flächen befinden. Mit der Annahme, dass sich ein Gebäude aus ebenen Flächen zusammensetzt, dient die Identifizierung von Clustern im Parameterraum der Detektierung dieser Flächen. Um diese Gruppen/Cluster aufzufinden wurde eine Clusteranalysetechnik genutzt. Über die detektierten Cluster können jene Laserscannerpunkte im Objektraum bestimmt werden, die eine Dachfläche formen. In die Laserscannerpunkte der somit gefundenen Dachflächen werden Ebenen interpoliert. Alle abgeleiteten Ebenen gehen in den entwickelten Rekonstruktionsalgorithmus ein, der eine Topologie zwischen den einzelnen Ebenen aufbaut. Anhand dieser Topologie erhalten die Ebenen ?Kenntnis? über ihre jeweiligen Nachbarn und können miteinander verschnitten werden. Der fertigen Dachgestalt werden Wände zugefügt und das komplette 3D Gebäudemodell wird mittels VRML (Virtual Reality Macro Language) visualisiert. Diese Studie bezieht sich neben der Entwicklung eines Schemas zu automatischen Gebäuderekonstruktion auch auf die Ableitung von Attributen der 3D Gebäudemodellen. Die entwickelte Methode wurde an verschiedenen Flugzeuglaserscannerdatensätzen getestet. Es wird gezeigt, welche Potentiale und Grenzen die entwickelte Methode bei der Bearbeitung dieser verschiedenen Laserscannerdatensätze hat.

Estimating large covariance matrices from small samples is an important problem in many fields. Among others, this includes spatial statistics and data assimilation. In this thesis, we deal with several methods of covariance estimation with emphasis on regula- rization and covariance models useful in filtering problems. We prove several properties of estimators and propose a new filtering method. After a brief summary of basic esti- mating methods used in data assimilation, the attention is shifted to covariance models. We show a distinct type of hierarchy in nested models applied to the spectral diagonal covariance matrix: explicit estimators of parameters are computed by the maximum like- lihood method and asymptotic variance of these estimators is shown to decrease when the maximization is restricted to a subspace that contains the true parameter value. A similar result is obtained for general M-estimators. For more complex covariance mo- dels, maximum likelihood method cannot provide explicit parameter estimates. In the case of a linear model for a precision matrix, however, consistent estimator in a closed form can be computed by the score matching method. Modelling of the precision ma- trix is particularly beneficial in Gaussian Markov random fields (GMRF), which possess a sparse precision matrix. The...