Submitted by Geyciane Santos (geyciane_thamires@hotmail.com) on 2015-06-18T14:37:14Z No. of bitstreams: 1 Dissertação - Felipe Gomes de Oliveira.pdf: 8063130 bytes, checksum: c3a3271d66779ae934aca4f3e28b999b (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-06-19T21:07:20Z (GMT) No. of bitstreams: 1 Dissertação - Felipe Gomes de Oliveira.pdf: 8063130 bytes, checksum: c3a3271d66779ae934aca4f3e28b999b (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-06-19T21:10:24Z (GMT) No. of bitstreams: 1 Dissertação - Felipe Gomes de Oliveira.pdf: 8063130 bytes, checksum: c3a3271d66779ae934aca4f3e28b999b (MD5) / Made available in DSpace on 2015-06-19T21:10:24Z (GMT). No. of bitstreams: 1 Dissertação - Felipe Gomes de Oliveira.pdf: 8063130 bytes, checksum: c3a3271d66779ae934aca4f3e28b999b (MD5) Previous issue date: 2011-07-25 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / This research presents a visual data fusion approach to recover dense depth map from sequences of images. The conventional methods to estimate depth map have many drawbacks with respect to environment illumination changes and camera positioning. We propose a Global optimization data fusion strategy to improve the measurements from stereo and focus depth maps. Di erent from typical stereo and focus fusion techniques, we use a single pair of stereo cameras to acquire series of images scenes without occlusion and illumination constraints. Then, we use Energy Functional fusion to associate the geometric coherence with multiple frames. In order to evaluate the results we de ned a metric using similarity measurements between traditional stereo and the proposed approach. The experiments are performed in real scene images, and the estimated mapping was superior than those found using traditional stereo methods, which demonstrates the good performance and robustness of our approach. / Este trabalho propõe uma abordagem de Fusão de Dados Visuais para estimar a estrutura tridimensional de uma cena a partir de sequências de imagens obtidas por meio de duas ou mais câmeras. Os métodos convencionais para estimar mapas de profundidade apresentam desvantagens relacionadas a mudanças na iluminação do ambiente e posicionamento de câmeras. Por essa razão, foi proposta uma estrategia de Fusão de Dados baseada em minimiza c~ao de energia para aprimorar as medições proporcionadas pela disparidade entre pixels de uma imagem e pela variação de foco. A abordagem proposta faz uso de uma rede distribuída de sensores visuais utilizando um par de câmeras estéreo sem restrições de oclusão ou iluminação no processo de captura de imagens. A função de energia foi usada para integrar múltiplos frames e inferir a coerência geométrica contida na cena. Para avaliar os resultados obtidos foram utilizadas métricas da literatura através de medições de similaridade entre técnicas de estéreo tradicionais e a estrategia desenvolvida. Os experimentos foram conduzidos a partir de imagens de cenas reais, e as informações de profundidade estimadas foram qualitativamente superior que os resultados obtidos pelos métodos tradicionais. Tais informações demonstram a qualidade dos resultados alcançados pela técnica proposta.

Desde a definção do problema de obtenção de regras de associação, vários algoritmos eficientes foram introduzidos para tratá-lo. Entretanto, ainda hoje o problema apresenta várias dificuldades práticas para os mineradores, como a determinação de limiares adequados de suporte mínimo e confiança mínima, a manipulação de grandes conjuntos de regras, e a compreensão de regras (especialmente aquelas contendo muitos itens). Para tratar estes problemas, pesquisadores têm investigado a aplicação de técnicas interativas, sumarização (de conjuntos de regras) e representações visuais. Entretanto, nenhuma abordagem na qual os usuários podem entender e controlar o processo por meio da interação com o algoritmo analítico ao longo de sua execução foi introduzida. Neste trabalho, é introduzida uma abordagem interativa para extração e exploração de regras de associação que insere o usuário no processo por meio de: execução interativa do Apriori ; seleção interativa de itemsets freqüentes; extração de regras baseada em itemsets e orientada por agrupamentos de itemsets similares; e exploração de regras aos pares. Para validar a abordagem, foram realizados diversos estudos, apoiados pelo Sistema \'I IND.2\' E, com o objetivo de: comparar a abordagem interativa, sob diversos aspectos, com uma abordagem convencional de obtenção de regras de associação; avaliar o efeito de variar alguns parâmetros do processo nos resultados finais; e mostrar a aplicação dos recursos oferecidos em situações reais e com usuários reais. Os resultados indicam que a abordagem apresentada é adequada, tanto em cenários exploratórios quanto em cenários em que há um direcionamento inicial para o processo, à execução de certas tarefas de extração de regras de associação, pois: provém recursos capazes de evitar execuções inteiras do algoritmo antes que os resultados sejam analisados; gera conjuntos de regras mais compactos; preserva a cobertura de itemsets; favorece a reformulação de tarefas ou a formulação de novas tarefas; e provê meios para comparação visual de regras, aumentando o poder de análise do minerador / Since the definition of the association rule mining problem, many efficient algorithms have been introduced to deal with it. However, the problem still presents many practical difficulties to the miners, such as the determination of suitable minimum support and minimum confidence thresholds, manipulation of large rule sets, and comprehension of rules (specially those containing many items). In order to deal with these problems, researchers have been investigating the application of interactive techniques, sumarization (of rule sets) and visual representations. Nonetheless, no approach in which users can understand and control the process through interaction with the analytical algorithm along its execution has been introduced. We introduce an interactive approach to extract and explore association rules that inserts the user into the process through: interactive execution of the Apriori ; interactive selection of frequent itemsets; itemset-based and cluster-oriented extraction of rules; and pairwise exploration of rules. To validate the approach, several studies have been conducted, supported by the \'I IND.2\' E System, aiming at: comparing the interactive approach, under several aspects, with a conventional approach to obtain association rules; evaluate the effect of different execution parameters in the final results; and illustrate its application in real situations and with real users. Results of these studies indicate that the approach is adequate, both in exploratory scenarios and in scenarios in which there is an initial guidance for the process, to the execution of certain association rule extraction tasks, because: it provides resources to avoid complete algorithm executions before results are analyzed; generates more compact rule sets for exploration; preserves rule diversity; favors the reformulation of tasks; and provides support for rule comparison, enhancing analysis capability for miners

Both human perceptual studies and automaticmachine-based experiments have shown that visual information from a speaker's mouth region can improve the robustness of automatic speech processing tasks, especially in the presence of acoustic noise. By taking advantage of the complementary nature of the acoustic and visual speech information, audio-visual speech processing (AVSP) applications can work reliably in more real-world situations than would be possible with traditional acoustic speech processing applications. The two most prominent applications of AVSP for viable human-computer-interfaces involve the recognition of the speech events themselves, and the recognition of speaker's identities based upon their speech. However, while these two fields of speech and speaker recognition are closely related, there has been little systematic comparison of the two tasks under similar conditions in the existing literature. Accordingly, the primary focus of this thesis is to compare the suitability of general AVSP techniques for speech or speaker recognition, with a particular focus on synchronous hidden Markov models (SHMMs). The cascading appearance-based approach to visual speech feature extraction has been shown to work well in removing irrelevant static information from the lip region to greatly improve visual speech recognition performance. This thesis demonstrates that these dynamic visual speech features also provide for an improvement in speaker recognition, showing that speakers can be visually recognised by how they speak, in addition to their appearance alone. This thesis investigates a number of novel techniques for training and decoding of SHMMs that improve the audio-visual speech modelling ability of the SHMM approach over the existing state-of-the-art joint-training technique. Novel experiments are conducted within to demonstrate that the reliability of the two streams during training is of little importance to the final performance of the SHMM. Additionally, two novel techniques of normalising the acoustic and visual state classifiers within the SHMM structure are demonstrated for AVSP. Fused hidden Markov model (FHMM) adaptation is introduced as a novel method of adapting SHMMs from existing wellperforming acoustic hidden Markovmodels (HMMs). This technique is demonstrated to provide improved audio-visualmodelling over the jointly-trained SHMMapproach at all levels of acoustic noise for the recognition of audio-visual speech events. However, the close coupling of the SHMM approach will be shown to be less useful for speaker recognition, where a late integration approach is demonstrated to be superior.

Nous nous intéressons dans cette thèse au problème de l’automatisation du processus de choix et de paramétrage des visualisations en fouille visuelle de données. Pour résoudre ce problème, nous avons développé un assistant utilisateur "VizAssist" dont l’objectif principal est de guider les utilisateurs (experts ou novices) durant le processus d’exploration et d’analyse de leur ensemble de données. Nous illustrons, l’approche sur laquelle s’appuie VizAssit pour guider les utilisateurs dans le choix et le paramétrage des visualisations. VizAssist propose un processus en deux étapes. La première étape consiste à recueillir les objectifs annoncés par l’utilisateur ainsi que la description de son jeu de données à visualiser, pour lui proposer un sous ensemble de visualisations candidates pour le représenter. Dans cette phase, VizAssist suggère différents appariements entre la base de données à visualiser et les visualisations qu’il gère. La seconde étape permet d’affiner les différents paramétrages suggérés par le système. Dans cette phase, VizAssist utilise un algorithme génétique interactif qui a pour apport de permettre aux utilisateurs d’évaluer et d’ajuster visuellement ces paramétrages. Nous présentons enfin les résultats de l’évaluation utilisateur que nous avons réalisé ainsi que les apports de notre outil à accomplir quelques tâches de fouille de données. / In this thesis, we deal with the problem of automating the process of choosing an appropriate visualization and its parameters in the context of visual data mining. To solve this problem, we developed a user assistant "VizAssist" which mainly assist users (experts and novices) during the process of exploration and analysis of their dataset. We illustrate the approach used by VizAssit to help users in the visualization selection and parameterization process. VizAssist proposes a process based on two steps. In the first step, VizAssist collects the user’s objectives and the description of his dataset, and then proposes a subset of candidate visualizations to represent them. In this step, VizAssist suggests a different mapping between the database for representation and the set of visualizations it manages. The second step allows user to adjust the different mappings suggested by the system. In this step, VizAssist uses an interactive genetic algorithm to allow users to visually evaluate and adjust such mappings. We present finally the results that we have obtained during the user evaluation that we performed and the contributions of our tool to accomplish some tasks of data mining.

Digitization that has enabled instant access to vast numbers of archival, historical images, demands a new paradigm for the use of visual imagery in historical research. This thesis proposes a new form of historiography in the digital medium, an image-based narrative mode for authoring and reading history. I propose a digital model for conveying history through the visual record, as an alternative to the printed book. Unlike the quantitative “big data” approach to digital humanities, this research explores visuality itself. In a practice-led approach, the research addresses both aspects of historiography: (1) a method of historical representation; and (2) original historical work on a selected topic. The testbed for historiographic and narrative experiments which led to the model was my case study on the Boxer Uprising in China, c. 1900. While many written histories of the Boxer Uprising exist, I collected a large portion of its extensive visual record for the first time. Sources from around the world, in a variety of media, were assembled into a digital data set that reveals previously unexplored historical themes. A series of visual narratives built in the case study culminated in a proposed “Visual Narrative Field” model. In this model, meaning emerges in the patterns observed between images within a complex visual field. The model vertically integrates three narrative approaches in order to support alternating cognitive modes used to read texts and perceive images. Linear concentration is blended with the non-linear exploration of interactive forms. The model provides historians with a much-needed tool for authoring narrative through relationships between images in a scalable approach. Due to digitization, visual databases are easily assembled, and images are as easily reproduced as written text. The Visual Narrative Field model takes advantage of the characteristics of the newly-digitized visual record, providing a means of authoring visual narrative that can be comprehended without the use of extensive written text. The model thus creates an unprecedented image-based method for performing and presenting historical research.

ICTs have enabled increased mobility and created a new era of workplace connectivity. Due to changes in work organization, global operations, increased mobility, and the new opportunities they are creating; work requires more coordination, more travel and a higher frequency of boundary spanning. ICTs have infiltrated into the personal life of individuals, while similarly, having an increasing impact on how organizations manage their workers‘ work-life balance. This research investigates the work boundary negotiation process in ICT-enabled work.Using an in-depth case study supplemented with visual data, this thesis studies the case of Sigma, an international consulting firm, that serves clients located in a large geographical area. It explores how consultants exhibiting mobile work practices, use ICTs to negotiate work boundaries. It draws on the structurational model of technology and complements it with the boundary object construct. The utilisation of this combined approach allows further understanding of work boundary negotiation.The research reveals that some ICTs as technological artefacts are boundary objects bridging between different groups of actors, crossing work boundaries, and allowing actors to negotiate their work boundaries while challenging traditional boundaries. Thus, allowing consultants to use their ICTs (specifically their smartphone) to negotiate their work boundaries on a need to basis. The boundary negotiation process (as revealed by the structuration process) is the means by which consultants try to make the most out of existing social structures – in this case specifically domination – in their organizational context. The ICT becomes a source of power and is mainly used to manifest domination over available resources. Consultants use them to maintain control over their life, increase their legitimacy and convey that they are professional experts. ICTs allow consultants to continuously redefine their work boundaries which become dynamic, fluid and contextual; the research reaffirms the sociotechnical nature of work boundaries.The thesis also develops a conceptual model of work boundary negotiation that conceptually illustrates how boundary negotiation is the outcome of the structuration process and the negotiation of existing structures of domination, legitimation and signification.

Repositórios com volumes de dados cada vez maiores foram viabilizados pelo desenvolvimento tecnológico, criando importantes fontes de informação em diversas áreas da atividade humana. Esses repositórios freqüentemente incluem informação sobre o comportamento temporal e o posicionamento espacial dos itens neles representados, os quais são extremamente relevantes para a análise dos dados. O processo de descoberta de conhecimento a partir de grandes volumes de dados tem sido objeto de estudo em diversas disciplinas, dentre elas a Visualização de Informação, cujas técnicas podem apoiar diversas etapas desse processo. Esta tese versa sobre o uso da Visualização Exploratória em conjuntos de dados com atributos temporais e espaciais, empregando a estratégia de múltiplas visualizações coordenadas para apoiar o tratamento de dados em estágios iniciais de processos de descoberta de conhecimento. São propostas duas novas representações visuais temporais – denominadas ‘Variação Temporal Uni-escala’ e ‘Variação Temporal Multi-escala’ – para apoiar a análise exploratória de dados temporais. Adicionalmente, é proposto um modelo de arquitetura de software – AdaptaVis, que permite a integração dessas e outras representações visuais em uma plataforma de visualização de informação flexível, extensível e adaptável às necessidades de diferentes usuários, tarefas e domínios de aplicação – a plataforma InfoVis. Sessões de uso realizadas com dados e usuários reais dos domínios de Climatologia e Negócios permitiram validar empiricamente as representações visuais e o modelo. O modelo AdaptaVis e a plataforma InfoVis estabelecem bases para a continuidade de diversas pesquisas em Visualização de Informação, particularmente o estudo de aspectos relacionados ao uso coordenado de múltiplas visualizações, à modelagem do processo de coordenação, e à integração entre múltiplas técnicas visuais e analíticas. / Data repositories with ever increasing volumes have been made possible by the evolution in data collection technologies, creating important sources of information in several fields of human activity. Such data repositories often include information about both the temporal behavior and the spatial positioning of data items that will be relevant in future data analysis tasks. The process of discovering knowledge embedded in great volumes of data is a topic of study in several disciplines, including Information Visualization, which offers a range of techniques to support different stages of a discovery process. This thesis addresses the application of Exploratory Visualization techniques on datasets with temporal and spatial attributes, using the strategy of coordinating multiple data views, to assist data treatment on early stages of knowledge discovery processes. Two temporal visual representations are proposed – ‘Uni-scale Temporal Behavior’ and ‘Multi-scale Temporal Behavior’ – that support the exploratory analysis of temporal data. Moreover, a software architecture model is introduced – AdaptaVis, that allows the integration of these and other visualization techniques into a flexible, extensible and adaptable information visualization platform – called InfoVis – that may be tailored to meet the requirements of different users, tasks and application domains. Sessions conducted with real data and users from the Climatology and Business application domains allowed an empirical validation of both the visual representations and the model. The AdaptaVis model and the InfoVis platform establish the basis for further research on issues related to the coordinated use of multiple data views, the modeling of the coordination process and the integration amongst multiple visual and analytical techniques.

Representações visuais têm sido adotadas na exploração de conjuntos de documentos, auxiliando a extração de conhecimento sem que seja necessária a análise individual de milhares de textos. Mapas de documentos, em particular, apresentam documentos individualmente representados espalhados em um espaço visual, refletindo suas relações de similaridade ou conexões. A construção destes mapas de documentos inclui, entre outras tarefas, o posicionamento dos textos e a identificação automática de áreas temáticas. Um desafio é a visualização de conjuntos dinâmicos de documentos. Na visualização de informação, é comum que alterações no conjunto de dados tenham um forte impacto na organização do espaço visual, dificultando a manutenção, por parte do usuário, de um mapa mental que o auxilie na interpretação dos dados apresentados e no acompanhamento das mudanças sofridas pelo conjunto de dados. Esta tese introduz um algoritmo para a construção dinâmica de mapas de documentos, capaz de manter uma disposição coerente à medida que elementos são adicionados ou removidos. O processo, inerentemente incremental e de baixa complexidade, utiliza um espaço bidimensional dividido em células, análogo a um tabuleiro de xadrez. Resultados consistentes foram alcançados em comparação com técnicas não incrementais de projeção de dados multidimensionais, tendo sido a técnica aplicada também em outros domínios, além de conjuntos de documentos. A visualização resultante não está sujeita a problemas de oclusão. A identificação de áreas temáticas é alcançada com técnicas de extração de regras de associação representativas para a identificação automática de tópicos. A combinação da extração de tópicos com a projeção incremental de dados em um processo integrado de mineração visual de textos compõe um espaço visual em que tópicos e áreas de interesse são destacados e atualizados à medida que o conjunto de dados é modificado / Visual representations are often adopted to explore document collections, assisting in knowledge extraction, and avoiding the thorough analysis of thousands of documents. Document maps present individual documents in visual spaces in such a way that their placement reflects similarity relations or connections between them. Building these maps requires, among other tasks, placing each document and identifying interesting areas or subsets. A current challenge is to visualize dynamic data sets. In Information Visualization, adding and removing data elements can strongly impact the underlying visual space. That can prevent a user from preserving a mental map that could assist her/him on understanding the content of a growing collection of documents or tracking changes on the underlying data set. This thesis presents a novel algorithm to create dynamic document maps, capable of maintaining a coherent disposition of elements, even for completely renewed sets. The process is inherently incremental, has low complexity and places elements on a 2D grid, analogous to a chess board. Consistent results were obtained as compared to (non-incremental) multidimensional scaling solutions, even when applied to visualizing domains other than document collections. Moreover, the corresponding visualization is not susceptible to occlusion. To assist users in indentifying interesting subsets, a topic extraction technique based on association rule mining was also developed. Together, they create a visual space where topics and interesting subsets are highlighted and constantly updated as the data set changes

This thesis focuses on advanced methods of detecting contours of the cardiac cells and measuring their contraction. The theoretical section describes the types of confocal microscopes, which are used for capturing biological samples. The following chapter is devoted to the methods of cardiac cells segmentation, where we are introduced to the generally applied approaches. The most widely spread methods of segmentation are active contours and mathematical morphology, which are the crucial topics of this thesis. Thanks to the those methods we are able in the visual data to accurately detect required elements and measure their surface chnage in time. Acquired theoretical knowledge leads us to the practical realization of the methods in MATLAB.

[pt] Um número crescente de sistemas de visualização tem sido desenvolvido tanto comercialmente quanto na comunidade de pesquisa. Embora estas ferramentas possam ajudar na construção de gráficos, elas apresentam desafios para analistas não especialistas. Um desafio em particular é o de prover suporte para destacar visualmente fatos de dados em gráficos. O esforço empregado por analistas não especialistas ou designers (sem conhecimento de programação) para realizar anotações visuais pode ser complexo e demorado. Nesta pesquisa, investigamos representações visuais de fatos de dados para apoiar analistas não especialistas na exploração e comunicação de insights através dos dados. Para endereçar estes desafios, nós tornamos operacional um modelo conceitual que relaciona visualizações, fatos de dados e suas representações visuais. Implementamos o modelo em uma ferramenta de visualização chamada VisStoryMaker, que permite gerar gráficos anotados sem exigir conhecimento especializado. Para avaliar o seu valor percebido, conduzimos um estudo de métodos mistos com usuário comparando com o Tableau Public. No geral, a VisStoryMaker oferece uma abordagem fácil de usar para destacar visualmente fatos sobre dados, e o uso de anotações visuais de fatos sobre dados nas visualizações podem apoiar usuários não especialistas na exploração e comunicação por meio de dados. Entretanto, seu uso deve ser cuidadosamente considerado para evitar poluir visualmente os gráficos. / [en] A growing number of visualization systems have been developed both commercially and within the research community. While these tools can aid in building charts, they can also present challenges for non-expert analysts. One particular challenge is providing support to visually highlight data facts in graphs. The manual effort employed by non-expert analysts or designers (without programming skills) to create annotations can be complex and time-consuming. In this research, we investigate visual representations of data facts in supporting non-expert analysts to explore and communicate insights through data. To address these challenges, we developed a conceptual model relating visualizations, data facts, and their visual representations. We implemented it into a visualization tool named VisStoryMaker, which allows generating annotated charts without requiring specialized knowledge. To benchmark its perceived value, we conducted a mixed-methods user study comparing it to Tableau Public. Overall, VisStoryMaker provides an easy-to-use approach to highlight facts visually, and the use of visual annotations in data visualizations can support non-expert users in data exploration and communication. However, their use must be carefully considered and designed to avoid visually cluttering the charts.