Imagens codificadas por matrizes de intensidade são tipicamente representadas por grande quantidade de dados. Embora existam inúmeras abordagens para análise de imagens, o conhecimento sobre problemas específicos é raramente considerado. Este trabalho trata sobre problemas de análises de imagens cujas soluções dependem do conhecimento sobre os dados envolvidos na aplicação específica. Para isso, utiliza técnicas de mineração de dados para modelar as respostas humanas obtidas de experimentos psicofísicos. Dois problemas de análise de imagens são apresentados: (1) a análise de formas e (2) a análise pictórica. No primeiro problema (1), formas de neurônios da retina (neurônios ganglionares de gato) são segmentadas e seus contornos submetidos a uma calibração dos parâmetros de curvatura considerando a segmentação manual de um especialista. Outros descritores, tais como esqueletos multi-escalas são explorados para eventual uso e avaliação da abordagem. No segundo problema (2), a análise pictórica de imagens de home-pages serve para avaliar critérios estéticos a partir de medidas de complexidade, contraste e textura. O sistema generaliza as respostas por um experimento psicofísico realizados com humanos. Os resultados objetivos com as duas abordagens revelaram-se promissores, surpreendentes e com ampla aplicabilidade. / Images coded by intensity matrices typically involve large amount of data. Although image analysis approaches are diverse, knowledge about specific problems is rarely considered. This work is about image analysis problems whose solutions depend on the knowledge about the involved data. In order to do so data mining techniques are applied to model human response to psychophysical experiments. Two image analysis problems are addressed: (1) shape analysis; and (2) pictorial analysis. In the former, neuronal images (ganglion retinal cells of cat) are segmented and curvature parameters are calibrated to identify extremities and branches on the shape considering human segmentation as a reference. Descriptors such as multiscale skeletons are also explored for potential application or evaluations. In the second problem, a pictorial analysis of home-pages images feed an artificial aesthetics criteria evaluator based on complexity, contrast and texture features. The system models and generalizes the obtained human responses to psychophysical experiment. The results for these two approaches are promising, surprising and widely applicable.

L’un des problèmes majeurs en analyse de formes est celui de l’analyse statistique en présence de formes aberrantes. On assiste avec l’évolution des moyens de collecte automatique des données, à la présence des valeurs aberrantes qui peuvent affecter énormément l’analyse descriptive des formes. En effet, les approches de l’état de l’art ne sont pas assez robustes à la présence de formes aberrantes. En particulier, la forme moyenne calculée penche vers les observations aberrantes et peut ainsi porter des déformations irrégulières. Aussi, l’analyse par ACP de la variabilité dans une classe de formes donnée conduit à des modes de variation qui décrivent plutôt la variabilité portée par ces formes aberrantes. Dans ce travail de thèse, nous proposons un schéma d’analyse robuste aux aberrations qui peuvent entacher une classe de formes donnée. Notre approche est une variante robuste de l’ACP qui consiste à détecter et à restaurer les formes aberrantes préalablement à une ACP menée dans l’espace tangent relatif à la forme moyenne. Au lieu de simplement éliminer les formes aberrantes, nous voulons bénéficier de la variabilité locale correcte qui y est présente en intégrant leur version restaurée dans l’analyse. Nous proposons également une approche variationnelle et une ACP élastique pour l’analyse de la variabilité d’un ensemble de formes en s’appuyant sur une métrique robuste basée géodésique. La troisième contribution de la thèse se situe au niveau des algorithmes de classification des formes basée sur les statistiques de formes : classification utilisant la moyenne intrinsèque, ou relaxée, par ACP tangente et par formes propres.Les approches proposées sont évaluées et comparées aux approches de l’état de l’art sur les bases de formes HAND et MPEG-7. Les résultats obtenus démontrent la capacité du schéma proposé à surpasser la présence de formes aberrantes et fournir des modes de variation qui caractérisent la variabilité des formes étudiées. / A major and complex problem in shape analysis is the statistical analysis of a set of shapes containing aberrant shapes. With the evolution of automatic data acquisition means, outliers can occur and their presence may greatly affect the descriptive analysis of shapes.Actually, state-of-the-art approaches are not robust enough to outliers. In particular, the calculated mean shape deviates towards the aberrant observations and thus carries irregular deformations.Similarly, the PCA analysis of the variability in a given class of shapes leads to variation modes which rather describe the variability carried by these aberrant shapes.In this thesis work, we propose a robust analysis scheme to handle the effects of aberrations that can occur in a given set. Our approach is a robust variant of PCA that consists in detecting and restoring aberrant shapes prior to a PCA in the tangent space relative to the means shape.Instead of simply rejecting outliers, we want to benefit from the present correct local variability by integrating their restored version into the analysis. We also propose a variational approach and an elastic PCA for the analysis of the variability of a set of shapes by using a robust geodesic-based metric. The third contribution of the thesis lies in the algorithms of shape classification based on shapes statistics: classification using the intrinsic mean shape, or relaxed one, by tangent PCA and by eigenshapes.The proposed schemes are evaluated and compared with existing schemes through two shape databases, HAND and MPEG-7. The results show the proposed scheme’s ability to overcome the presence of aberrant shapes and provide variation modes that characterize the variability of studied shapes.

Dans le domaine moléculaire, les polymères sont observés et étudiés par microscopie. Les formes obtenues sont souvent imprécises en raison de l'effet de convolution et de diffraction de l'acquisition par microscopie. Par conséquent, le polymère apparait comme une courbe épaisse, bruitée et floue. Pour étudier les caractéristiques d'une chaîne de polymère, une des approches possibles consiste à réduire la forme acquise par microscope à une représentation minimale, soit une courbe. Cette dernière doit représenter au mieux l'objet étudié malgré les différentes difficultés rencontrées telles que la qualité des images ou les imprécisions dues à la discrétisation. De plus, un polymère adopte un mouvement "Reptilien" et peut former des géométries complexes telles que des courbes fermées ou avec boucles. L'objet de cette thèse est donc l'extraction de courbes visant à fournir une représentation minimale des polymères à des fins d'analyse. La méthode que nous proposons comprend deux grandes étapes à savoir : L'extraction des géodésiques et leur fusion. La première étape consiste à calculer un ensemble de géodésiques, chacune parcourant une partie distincte de la forme. Ces morceaux de géodésiques sont fusionnés dans la seconde étape afin de générer la courbe complète. Afin de représenter la reptation, les géodésiques doivent être fusionnées dans un ordre précis. Nous modélisons ce problème par graphes et nous cherchons l'ordre de fusion en parcourant le graphe. La fusion est effectuée selon le chemin optimal minimisant différentes contraintes. / In the molecular field, polymers are observed and studied by microscopy. The shapes obtained are often inaccurate because of the convolution and diffraction effect of microscopy. Therefore, the polymer appears as a thick, noisy and fuzzy curve. In order to study a polymer chain, one of the possible approaches consists in reducing the acquired shape to a minimal representation, ie a curve. This curve must represent the studied object in the best way, despite the various encountered difficulties, such as the quality of the images or inaccuracies due to discretization. In addition, a polymer performs a "Reptilian" movement and can form complex geometries such as closed and looped curves. The object of this thesis is, therefore, the extraction of curves aimed at providing a minimal representation of the polymers for analysis. The proposed method comprises two major steps: geodesics extraction and their fusion. The first step is to compute a set of geodesics, each one traversing a distinct part of the shape. These pieces of geodesics are fused in the second step to generate the complete curve. In order to represent the reptation, the geodesics have to be merged in a precise order. We model this problem by graphs and consider the fusion as a graph traversal problem. The fusion is performed according to the optimal path minimizing various constraints.

There is an understanding among biological anthropologists that long bone epiphyseal shape is highly regulated by genetic and biomechanical factors. Conversely, long bone diaphyseal geometry and robusticity have been shown to respond to activity in life. The current study examined the assumption of epiphyseal consistency by exploring the relationship between a well established bony response to activity (Cross-Sectional Geometry) and shape change among the proximal humerus and femur. Long bone samples were taken from four hunter-gatherer populations: the Andaman Islanders, the Indian Knoll, Point Hope Alaskans, and the Sadlermiut. Shape was measured through landmark configurations placed on the proximal end of a total of 91 humeri and 84 femora. Cross-sectional Geometry measures (J) were taken from each specimen, as well. Principal Component Analyses were conducted on the landmark shape data to determine where the shape variation was occurring among the sample. These Principal Components were then compared via Bivariate Regression to the J values taken from the diaphysis. Significant relationships occurred between the development of the lesser tubercle and an increase in J among the humerus sample. Significant relationships were also found among the femur sample; as when J increased the proximal epiphyses were more likely to be more gracile, and the space between the femoral head and the greater trochanter increased. The humerus results indicated a more robust proximal epiphysis in groups with activities that rely heavily on the upper body, such as rowing, swimming, and harpooning. The femur results were more complex, as the relationship between activity and proximal shape is likely heavily influenced by a genetically predetermined body shape. These results indicated that there is a relationship between activity and proximal epiphyseal shape, but that it, like all relationships, is complex, and comprised of multiple factors. Ultimately, proximal long bone shape is the result of multiple influences including, activity, genetics, population adaptation, health, and many more. Future research should focus on determining if the relationship between activity and shape exists among other populations, and when and where it begins during growth and development. / Graduate

This paper describes how feature extraction on ancient pottery can be combined with recent developments in artificial intelligence to draw up an automated, but still flexible classification system. These features include for instance several dimensions of the vessel's body, ratios thereof, an abstract representation of the overall shape, the shape of vessel segments and the number and type of attachments such as handles, lugs and feet. While most traditional approaches to classification are based on statistical analysis or the search for fuzzy clusters in high-dimensional spaces, we apply machine learning techniques, such as decision tree algorithms and neural networks. These methods allow for an objective and reproducible classification process. Conclusions about the "typability" of data, the evolution of types and the diagnostic attributes of the types themselves can be drawn as well.

Les images de synthèse sont présentes à travers un grand nombre d'applications tel que les jeux vidéo, le cinéma, l'architecture, la publicité, l'art, la réalité virtuelle, la visualisation scientifique, l'ingénierie en éclairage, etc. En conséquence, la demande en photoréalisme et techniques de rendu rapide ne cesse d'augmenter. Le rendu réaliste d'une scène virtuelle nécessite l'estimation de son illumination globale grâce à une simulation du transport de lumière, un processus coûteux en temps de calcul dont la vitesse de convergence diminue généralement lorsque la complexité de la scène augmente. En particulier, une forte illumination indirecte combinée à de nombreuses occlusions constitue une caractéristique globale de la scène que les techniques existantes ont du mal à gérer. Cette thèse s'intéresse à ce problème à travers l'application de techniques d'analyse de formes pour le rendu 3D.Notre principal outil est un squelette curviligne du vide de la scène, représenté par un graphe contenant des informations sur la topologie et la géométrie de la scène. Ce squelette nous permet de proposer de nouvelles méthodes pour améliorer des techniques de rendu temps réel et non temps réel. Concernant le rendu temps réel, nous utilisons les informations géométriques du squelette afin d'approximer le rendu des ombres projetés par un grand nombre de points virtuels de lumière représentant l'illumination indirecte de la scène 3D.Pour ce qui est du rendu non temps réel, nos travaux se concentrent sur des algorithmes basés sur l'échantillonnage de chemins, constituant actuellement le principal paradigme en rendu physiquement plausible. Notre squelette mène au développement de nouvelles stratégies d'échantillonnage de chemins, guidés par des caractéristiques topologiques et géométriques. Nous adressons également ce problème à l'aide d'un second outil d'analyse de formes: la fonction d'ouverture du vide de la scène, décrivant l'épaisseur locale du vide en chacun de ses points. Nos contributions offrent une amélioration des méthodes existantes and indiquent clairement que l'analyse de formes offre de nombreuses opportunités pour le développement de nouvelles techniques de rendu 3D / Nowadays, computer generated images can be found everywhere, through a wide range of applications such as video games, cinema, architecture, publicity, artistic design, virtual reality, scientific visualization, lighting engineering, etc. Consequently, the need for visual realism and fast rendering is increasingly growing. Realistic rendering involves the estimation of global illumination through light transport simulation, a time consuming process for which the convergence rate generally decreases as the complexity of the input virtual 3D scene increases. In particular, occlusions and strong indirect illumination are global features of the scene that are difficult to handle efficiently with existing techniques. This thesis addresses this problem through the application of discrete shape analysis to rendering. Our main tool is a curvilinear skeleton of the empty space of the scene, a sparse graph containing important geometric and topological information about the structure of the scene. By taking advantage of this skeleton, we propose new methods to improve both real-time and off-line rendering methods. Concerning real-time rendering, we exploit geometric information carried by the skeleton for the approximation of shadows casted by a large set of virtual point lights representing the indirect illumination of the 3D scene. Regarding off-line rendering, our works focus on algorithms based on path sampling, that constitute the main paradigm of state-of-the-art methods addressing physically based rendering. Our skeleton leads to new efficient path sampling strategies guided by topological and geometric features. Addressing the same problem, we also propose a sampling strategy based on a second tool from discrete shape analysis: the opening function of the empty space of the scene, describing the local thickness of that space at each point. Our contributions demonstrate improvements over existing approaches and clearly indicate that discrete shape analysis offers many opportunities for the development of new rendering techniques

A complexidade é uma característica de grande importância em processos de reconhecimento de padrões, especialmente naqueles que envolvem imagens biológicas. Este trabalho tem como objetivo estudar métodos que realizam a análise de imagens por meio da análise de sua complexidade. Os métodos a serem estudados foram selecionados com base na similaridade de seus algoritmos e metodologia: dimensão fractal, Caminhada Determinística do Turista e Redes Complexas. Estes métodos permitem realizar a análise e segmentação de formas ou texturas contidas em uma imagem com base na sua variação de complexidade. Dos três métodos considerados, dois deles fazem parte do estado da arte em análise de complexidade, enquanto que a dimensão fractal já é aplicada a mais tempo na análise de formas e texturas. Os trabalhos aqui desenvolvidos visam comparar e analisar os métodos selecionados por meio de experimentos com imagens de forma e texturas, sendo utilizadas texturas naturais e de Brodatz, freqüentemente utilizadas na literatura como benchmark para texturas. Com base no conhecimento adquirido, novas técnicas voltadas para a análise e segmentação de formas e texturas foram desenvolvidas, assim como foram analisadas as deficiências e propostas melhorias às técnicas estudadas. Além disso, diversos experimentos com estas metodologias foram realizados em aplicações de Bioinformática / Complexity is a feature of great importance in pattern recognition processes, especially those involving biological images. This work aims to study methods that perform image analysis by the analysis of its complexity. The methods to be studied were selected based on similarity of their algorithms and methodology: fractal dimension, Deterministic Tourist Walk and Complex Networks. These methods enable us to perform the analysis and segmentation of shapes and textures contained in an image based on the variation of its complexity. Of the three methods considered, two of them are part of the state of the art in complexity analysis, while the fractal dimension is already applied in shapes and textures analysis. The work developed here aims to compare and analyze the selected methods through experiments with shape and texture images, utilizing for this natural and Brodatz textures samples, often used in literature as benchmark for textures analysis. Based on the knowledge acquired, new techniques for analysis and segmentation of shapes and textures were developed, as also were analyzed the deficiencies and proposed improvements to the techniques studied. Moreover, several experiments with these methods were performed in bioinformatics applications

O gênero Eimeria compreende um grupo de protozoários da classe Coccidia que infecta uma grande variedade de hospedeiros. Um total de sete espécies distintas Eimeria podem infectar a galinha doméstica causando enterites com graves prejuízos econômicos. A identificação das espécies pode ser feita através da análise microscópica das diferentes características morfológicas dos oocistos, um dos estágios de desenvolvimento do parasita. Alternativamente, ensaios moleculares baseados na amplificação de alvos específicos de DNA também podem ser utilizados. Em ambos os casos, requer-se um laboratório especializado e, principalmente, pessoal altamente treinado. Neste trabalho é relatada uma abordagem computacional para a extração automática de características para a representação da forma das distintas espécies de Eimeria. Foram utilizadas imagens digitais do protozoário nas quais aplicou-se técnicas de processamento de imagens e visão computacional para sua representação morfológica, formando três grupos de características: medidas geométricas, caracterização da curvatura, e quantificação da estrutura interna. A morfologia dos protozoários foi representada por um vetor de características constituído por 14 dimensões, o qual constituiu o padrão de entrada para o processo de classificação. Para o reconhecimento dos padrões, foram usados dois classificadores Bayesianos, utilizando-se como funções de verossimilhança a Gaussiana e a de Dirichlet, respectivamente. O primeiro classificador apresentou as melhores taxas de acerto, enquanto o segundo demonstrou melhor desempenho segundo a análise por curvas ROC. Como prova de princípio de que o sistema poderia ser utilizado por usuários leigos para o diagnóstico à distância de parasitas, foi implementado o COCCIMORPH, um sistema de diagnóstico de Eimeria em tempo real. O sistema permite o envio de imagens via web, assim como o seu pré-processamento e classificação remotos, obtendo-se o resultado do diagnóstico em tempo real. Essa abordagem totalmente integrada e implementada é inédita para o diagnóstico de parasitas. Entre suas vantagens principais está o fato de que o diagnóstico pode ser obtido sem a necessidade do transporte físico de amostras biológicas para um laboratório de referência, evitando assim riscos de contaminação do ambiente. Para o treinamento do sistema, foram obtidas centenas de micrografias de cada uma das sete espécies de Eimeria que infectam a galinha doméstica. Essas imagens também foram usadas para a construção de um banco de acesso público de imagens (The Eimeria Image Database). Além disso, a metodologia de diagnóstico foi também aplicada e testada com onze espécies Eimeria de coelho doméstico. Com isso, foram gerados dados inéditos de morfometria, micrografias adicionais para o banco de imagens, e um sistema de classificação para esse conjunto adicional de parasitas. Finalmente, foram determinadas as distâncias entre as diferentes espécies de Eimeria, calculadas a partir dos dados morfométricos. As árvores de distância revelaram uma topologia muito similar com árvores obtidas a partir da inferência filogenética usando-se marcadores moleculares como o gene 18S de rRNA ou genomas mitocondriais. / The Eimeria genus comprises a group of protozoan parasites that infect a wide range of hosts. A total of seven different Eimeria species infect the domestic fowl, causing enteritis with severe economical losses. Species identification can be performed through microscopic analysis of the distinct morphological characteristics of the oocysts, a developmental stage of the parasite. Alternatively, molecular assays based on the amplification of specific DNA targets can also be used. In both cases, a well equipped laboratory and, especially, highly qualified personnel are required. In this work, we report a computational approach for the automatic feature extraction for shape representation of the different Eimeria species. Digital images of the parasites were used in order to apply image processing and computational vision techniques for shape characterization. Three groups of morphological features were constituted: geometric measures, curvature characterization, and internal structure quantification. The protozoan morphology was represented by a 14-dimension feature vector, which was used as the input pattern for the classification process. Two Bayesian classifiers were used for pattern recognition, using as a likelihood function the normal and the Dirichlet, respectively. The former classifier presented the best correct classification rates, whereas the latter showed a better performance in ROC curve analyses. As a proof of principle that this system could be utilized by end-users for a long-distance parasite diagnosis, we implemented COCCIMORPH, an integrated system for the real-time diagnosis of Eimeria spp. The system presents an interface for image uploading. Image preprocessing and diagnosis are performed remotely and the results displayed in real-time. This fully integrated and implemented system constitutes a novel approach for parasite diagnosis. Among the several advantages of the system, it is noteworthy that no biological sample transportation is required between the farm and the reference laboratory, thus avoiding potential environment contamination risks. To train the system, we used hundreds of micrographs of each one of the seven Eimeria species of domestic fowl. These images were used to compose a public image repository (The Eimeria Image Database). In addition, our diagnosis methodology was extended to the eleven Eimeria species that infect the domestic rabbit. With this integrated approach, a totally novel set of images and morphometric data of rabbit Eimeria were incorporated to the image database and, also to the remote diagnosis system. Finally, distance trees of the distinct Eimeria species of domestic fowl were computed from the morphometric data. The trees revealed a very similar topology with trees obtained with molecular phylogenetic markers such as the 18S rRNA gene and mitochondrial genomes.

Dans cette thèse, nous développons un modèle du second ordre pour la représentation des formes (courbes et surfaces) grâce à la théorie des cycles normaux. Le cycle normal d'une forme est le courant associé à son fibré normal. En introduisant des métriques à noyaux sur les cycles normaux, nous obtenons une mesure de dissimilarité entre formes qui prend en compte leurs courbures. Cette mesure est ensuite utilisée comme terme d'attache aux données dans une optique d'appariement et d'analyse de formes par les déformations. Le chapitre 1 est une revue du domaine de l'analyse de formes par les déformations. Nous insistons plus particulièrement sur la mise en place théorique et numérique du modèle de Large Deformation Diffeomorphic Metric Mapping (LDDMM). Le chapitre 2 se concentre sur la représentation des formes par les cycles normaux dans un cadre unifié qui englobe à la fois les formes continues et discrètes. Nous précisons dans quelle mesure cette représentation contient des informations de courbure. Enfin nous montrons le lien entre le cycle normal d'une forme et son varifold. Dans le chapitre 3, nous introduisons les métriques à noyaux. Ainsi, nous pouvons considérer les cycles normaux dans un espace de Hilbert avec un produit scalaire explicite. Nous détaillons ce produit scalaire dans le cas des courbes et surfaces discrètes avec certains noyaux, ainsi que le gradient associé. Nous montrons enfin que malgré le choix de noyaux simples, nous ne perdons pas toutes les informations de courbures. Le chapitre 4 utilise cette nouvelle métrique comme terme d'attache aux données dans le cadre LDDMM. Nous présentons de nombreux appariements et estimations de formes moyennes avec des courbes ou des surfaces. L'objectif de ce chapitre est d'illustrer les différentes propriétés des cycles normaux pour l'analyse des déformations sur des exemples synthétiques et réels. / In this thesis, we develop a second order model for the representation of shapes (curves or surfaces) using the theory of normal cycles. The normal cycle of a shape is the current associated with its normal bundle. Introducing kernel metrics on normal cycles, we obtain a dissimilarity measure between shapes which takes into account curvature. This measure is used as a data attachment term for a purpose of registration and shape analysis by deformations. Chapter 1 is a review of the field of shape analysis. We focus on the setting of the theoretical and numerical model of the Large Deformation Diffeomorphic Metric Mapping(LDDMM).Chapter 2 focuses on the representation of shapes with normal cycles in a unified framework that encompasses both the continuous and the discrete shapes. We specify to what extend this representation encodes curvature information. Finally, we show the link between the normal cycle of a shape and its varifold. In chapter 3, we introduce the kernel metrics, so that we can consider normal cycles in a Hilbert space with an explicit scalar product. We detail this scalar product for discrete curves and surfaces with some kernels, as well as the associated gradient. We show that even with simple kernels, we do not get rid of all the curvature informations. The chapter 4 introduces this new metric as a data attachment term in the framework of LDDMM. We present numerous registrations and mean shape estimation for curves and surfaces. The aim of this chapter is to illustrate the different properties of normal cycles for the deformations analysis on synthetic and real examples.

This diploma thesis is focused on time-development analysis of treated lesion in CT data. The theoretical part of the thesis deals with the anatomy, physiology, and pathophysiology of the spine and vertebral bodies. It further describes diagnostic and therapeutic options for the detection and treatment of spinal lesions. It contains an overview of the current state of usage of time-development analysis in oncology. The problems of the available databases are discussed and new databases are created for subsequent analysis. Futhermore, the methodology of time-development analysis according to the shape characterization and the size of the vertebral involvement is proposed. The proposed methodological approaches to feature extraction are applied to the created databases. Their choice and suitability is discussed, including their potential for possible usege in clinical practice of monitoring the development and derivation of characteristic dependences of features on the patient's prognosis.