Segmentação de objetos em imagens é um dos problemas mais fundamentais e desafiadores em processamento de imagem e visão computacional. O conhecimento de alto nível e específico do usuário é frequentemente requerido no processo de segmentação, devido à presença de fundos heterogêneos, objetos com bordas fracamente definidas, inomogeneidade de campo, ruído, artefatos, efeitos de volume parcial e seus efeitos conjuntos. Propriedades globais do objeto de interesse, tais como conexidade, restrições de forma e polaridade de borda, são conhecimentos prévios de alto nível úteis para a sua segmentação, permitindo a customização da segmentação para um objeto alvo. Nesse trabalho, apresentamos um novo método chamado Transformada Imagem-Floresta Orientada Conexa (COIFT, Connected Oriented Image Foresting Transform), que fornece soluções ótimas globais de acordo com uma medida de corte em grafo, incorporando a restrição de conexidade na Transformada Imagem-Floresta Orientada (OIFT, Oriented Image Foresting Transform), com o fim de garantir a geração de objetos conexos, bem como permitir o controle simultâneo da polaridade de borda. Enquanto o emprego de restrições de conexidade em outros arcabouços, tais como no algoritmo de corte-mínimo/fluxo-máximo (min-cut/max-flow), leva a um problema NP-difícil, a COIFT conserva o baixo custo computacional da OIFT. Experimentos mostram que a COIFT pode melhorar consideravelmente a segmentação de objetos com partes finas e alongadas, para o mesmo número de sementes em segmentação baseada em marcadores. / Object segmentation is one of the most fundamental and challenging problems in image processing and computer vision. The high-level and specific knowledge of the user is often required in the segmentation process, due to the presence of heterogeneous backgrounds, objects with poorly defined boundaries, field inhomogeneity, noise, artifacts, partial volume effects and their joint effects. Global properties of the object of interest, such as connectivity, shape constraints and boundary polarity, are useful high-level priors for its segmentation, allowing the customization of the segmentation for a given target object. In this work, we introduce a new method called Connected Oriented Image Foresting Transform (COIFT), which provides global optimal solutions according to a graph-cut measure in graphs, subject to the connectivity constraint in the Oriented Image Foresting Transform (OIFT), in order to ensure the generation of connected objects, as well as allowing the simultaneous control of the boundary polarity. While the use of connectivity constraints in other frameworks, such as in the min-cut/max-flow algorithm, leads to a NP-Hard problem, COIFT retains the low computational cost of OIFT. Experiments show that COIFT can considerably improve the segmentation of objects with thin and elongated parts, for the same number of seeds in segmentation based on markers.

No description available.

Many different approaches have been taken towards solving the stereo correspondence problem and great progress has been made within the field during the last decade. This is mainly thanks to newly evolved global optimization techniques and better ways to compute pixel dissimilarity between views. The most successful algorithms are based on approaches that explicitly model smoothness assumptions made about the physical world, with image segmentation and plane fitting being two frequently used techniques. Within the project, a survey of state of the art stereo algorithms was conducted and the theory behind them is explained. Techniques found interesting were implemented for experimental trials and an algorithm aiming to achieve state of the art performance was implemented and evaluated. For several cases, state of the art performance was reached. To keep down the computational complexity, an algorithm relying on local winner-take-all optimization, image segmentation and plane fitting was compared against minimizing a global energy function formulated on pixel level. Experiments show that the local approach in several cases can match the global approach, but that problems sometimes arise – especially when large areas that lack texture are present. Such problematic areas are better handled by the explicit modeling of smoothness in global energy minimization. Lastly, disparity estimation for image sequences was explored and some ideas on how to use temporal information were implemented and tried. The ideas mainly relied on motion detection to determine parts that are static in a sequence of frames. Stereo correspondence for sequences is a rather new research field, and there is still a lot of work to be made.

Ce travail de thèse est axé sur le domaine du traitement d'images biomédicales. L'objectif de notre étude est l'estimation des paramètres traduisant les propriétés mécaniques de l'artère carotide in vivo en imagerie échographique, dans une optique de détection précoce de la pathologie cardiovasculaire. L'analyse du mouvement longitudinal des tissus de la paroi artérielle, i.e. dans la même direction que le flux sanguin, représente la motivation majeure de ce travail. Les trois contributions principales proposées dans ce travail sont i) le développement d'un cadre méthodologique original et semi-automatique, dédié à la segmentation et à l'estimation du mouvement de la paroi artérielle dans des séquences in vivo d'images ultrasonores mode-B, ii) la description d'un protocole de génération d'une référence, faisant intervenir les opérations manuelles de plusieurs experts, dans le but de quantifier la précision des résultats de notre méthode malgré l'absence de vérité terrain inhérente à la modalité échographique, et iii) l'évaluation clinique de l'association entre les paramètres mécaniques et dynamiques de la paroi carotidienne et les facteurs de risque cardiovasculaire dans le cadre de la détection précoce de l'athérosclérose. Nous proposons une méthode semi-automatique, basée sur une approche conjointe de segmentation des contours de la paroi et d'estimation du mouvement des tissus. L'extraction de la position des interfaces est réalisée via une approche spécifique à la structure morphologique de la carotide, basée sur une stratégie de programmation dynamique exploitant un filtrage adapté. L'estimation du mouvement est réalisée via une méthode robuste de mise en correspondance de blocs (block matching), basée sur la connaissance du déplacement a priori ainsi que sur la mise à jour temporelle du bloc de référence par un filtre de Kalman spécifique. La précision de notre méthode, évaluée in vivo, correspond au même ordre de grandeur que celle résultant des opérations manuelles réalisées par des experts, et reste sensiblement meilleure que celle obtenue avec deux autres méthodes traditionnelles (i.e. une implémentation classique de la technique de block matching et le logiciel commercial Velocity Vector Imaging). Nous présentons également quatre études cliniques réalisées en milieu hospitalier, où nous évaluons l'association entre le mouvement longitudinal et les facteurs de risque cardiovasculaire. Nous suggérons que le mouvement longitudinal, qui représente un marqueur de risque émergent et encore peu étudié, constitue un indice pertinent et complémentaire aux marqueurs traditionnels dans la caractérisation de la physiopathologie artérielle, reflète le niveau de risque cardiovasculaire global, et pourrait être bien adapté à la détection précoce de l'athérosclérose. / This thesis is focused on the domain of bio-medical image processing. The aim of our study is to assess in vivo the parameters traducing the mechanical properties of the carotid artery in ultrasound imaging, for early detection of cardiovascular diseases. The analysis of the longitudinal motion of the arterial wall tissues, i.e. in the same direction as the blood flow, represents the principal motivation of this work. The three main contributions proposed in this work are i) the development of an original and semi-automatic methodological framework, dedicated to the segmentation and motion estimation of the arterial wall in in vivo ultrasound B-mode image sequences, ii) the description of a protocol aiming to generate a reference, involving the manual tracings of several experts, in the objective to quantify the accuracy of the results of our method despite the absence of ground truth inherent to ultrasound imaging, and iii) the clinical evaluation of the association between the mechanical and dynamical parameters of the arterial wall and the cardiovascular risk factors, for early detection of atherosclerosis. We propose a semi-automatic method, based on a combined approach of wall segmentation and tissues motion estimation. The extraction on the interfaces position is realized via an approach specific to the morphological structure of the carotid artery, based on a strategy of dynamic programming using a matched filter. The motion estimation is performed via a robust block matching method, based on the a priori knowledge of the displacement as well as the temporal update of the reference block with a specific Kalman filter. The accuracy of our method, evaluated in vivo, corresponds to the same order of magnitude as the one resulting from the manual operations performed by experts, and is significantly higher than the one obtained from two other classical methods (i.e. a classical implementation of the block matching technique, and the VVI commercial software). We also present four clinical studies, and we evaluate the association between longitudinal motion and cardiovascular risk factors. We suggest that the longitudinal motion, which represents an emerging cardiovascular risk marker that has been only few studied yet, constitutes a pertinent and complementary marker aiming at the characterization of arterial physio-pathology, traduces the overall cardiovascular risk level, and could be well suited to the early detection of the atherosclerosis.

De plus en plus souvent, les études médicales utilisent simultanément de multiples modalités d'acquisition d'image, produisant ainsi des données multidimensionnelles comportant beaucoup d'information supplémentaire dont l'interprétation et le traitement deviennent délicat. Par exemple, les études sur l'ischémie cérébrale se basant sur la combinaison de plusieurs images IRM, provenant de différentes séquences d'acquisition, pour prédire l'évolution de la zone nécrosée, donnent de bien meilleurs résultats que celles basées sur une seule image. Ces approches nécessitent cependant l'utilisation d'algorithmes plus complexes pour réaliser les opérations de filtrage, segmentation et de clustering. Une approche robuste pour répondre à ces problèmes de traitements de données multidimensionnelles est le Mean Shift qui est basé sur l'analyse de l'espace des caractéristiques et l'estimation non-paramétrique par noyau de la densité de probabilité. Dans cette thèse, nous étudions les paramètres qui influencent les résultats du Mean Shift et nous cherchons à optimiser leur choix. Nous examinons notamment l'effet du bruit et du flou dans l'espace des caractéristiques et comment le Mean Shift doit être paramétrés pour être optimal pour le débruitage et la réduction du flou. Le grand succès du Mean Shift est principalement du au réglage intuitif de ces paramètres de la méthode. Ils représentent l'échelle à laquelle le Mean Shift analyse chacune des caractéristiques. En se basant sur la méthode du Plug In (PI) monodimensionnel, fréquemment utilisé pour le filtrage Mean Shift et permettant, dans le cadre de l'estimation non-paramétrique par noyau, d'approximer le paramètre d'échelle optimal, nous proposons l'utilisation du PI multidimensionnel pour le filtrage Mean Shift. Nous évaluons l'intérêt des matrices d'échelle diagonales et pleines calculées à partir des règles du PI sur des images de synthèses et naturelles. Enfin, nous proposons une méthode de segmentation automatique et volumique combinant le filtrage Mean Shift et la croissance de région ainsi qu'une optimisation basée sur les cartes de probabilité. Cette approche est d'abord étudiée sur des images IRM synthétisées. Des tests sur des données réelles issues d'études sur l'ischémie cérébrale chez le rats et l'humain sont aussi conduits pour déterminer l'efficacité de l'approche à prédire l'évolution de la zone de pénombre plusieurs jours après l'accident vasculaire et ce, à partir des IRM réalisées peu de temps après la survenue de cet accident. Par rapport aux segmentations manuelles réalisées des experts médicaux plusieurs jours après l'accident, les résultats obtenus par notre approche sont mitigés. Alors qu'une segmentation parfaite conduirait à un coefficient DICE de 1, le coefficient est de 0.8 pour l'étude chez le rat et de 0.53 pour l'étude sur l'homme. Toujours en utilisant le coefficient DICE, nous déterminons la combinaison de d'images IRM conduisant à la meilleure prédiction. / Medical studies increasingly use multi-modality imaging, producing multidimensional data that bring additional information that are also challenging to process and interpret. As an example, for predicting salvageable tissue, ischemic studies in which combinations of different multiple MRI imaging modalities (DWI, PWI) are used produced more conclusive results than studies made using a single modality. However, the multi-modality approach necessitates the use of more advanced algorithms to perform otherwise regular image processing tasks such as filtering, segmentation and clustering. A robust method for addressing the problems associated with processing data obtained from multi-modality imaging is Mean Shift which is based on feature space analysis and on non-parametric kernel density estimation and can be used for multi-dimensional filtering, segmentation and clustering. In this thesis, we sought to optimize the mean shift process by analyzing the factors that influence it and optimizing its parameters. We examine the effect of noise in processing the feature space and how Mean Shift can be tuned for optimal de-noising and also to reduce blurring. The large success of Mean Shift is mainly due to the intuitive tuning of bandwidth parameters which describe the scale at which features are analyzed. Based on univariate Plug-In (PI) bandwidth selectors of kernel density estimation, we propose the bandwidth matrix estimation method based on multi-variate PI for Mean Shift filtering. We study the interest of using diagonal and full bandwidth matrix with experiment on synthesized and natural images. We propose a new and automatic volume-based segmentation framework which combines Mean Shift filtering and Region Growing segmentation as well as Probability Map optimization. The framework is developed using synthesized MRI images as test data and yielded a perfect segmentation with DICE similarity measurement values reaching the highest value of 1. Testing is then extended to real MRI data obtained from animals and patients with the aim of predicting the evolution of the ischemic penumbra several days following the onset of ischemia using only information obtained from the very first scan. The results obtained are an average DICE of 0.8 for the animal MRI image scans and 0.53 for the patients MRI image scans; the reference images for both cases are manually segmented by a team of expert medical staff. In addition, the most relevant combination of parameters for the MRI modalities is determined.

Nous développons une méthode générique semi-automatique multi-objet de segmentation d'image par coupure de graphe visant les usages médicaux de routine, allant des tâches impliquant quelques objets dans des images 2D, à quelques dizaines dans celles 3D quasi corps entier. La formulation souple de la méthode permet son adaptation simple à une application donnée. En particulier, le modèle d'a priori de proximité que nous proposons, défini à partir des contraintes de paires du plus court chemin sur le graphe d'adjacence des objets, peut facilement être adapté pour tenir compte des relations spatiales entre les objets ciblés dans un problème donné. L'algorithme de segmentation peut être adapté aux besoins de l'application en termes de temps d'exécution et de capacité de stockage à l'aide d'une partition de l'image à segmenter par une tesselation de Voronoï efficace et contrôlable, établissant un bon équilibre entre la compacité des régions et le respect des frontières des objets. Des évaluations et comparaisons qualitatives et quantitatives avec le modèle de Potts standard confirment que notre modèle d'a priori apporte des améliorations significatives dans la segmentation d'objets distincts d'intensités similaires, dans le positionnement précis des frontières des objets ainsi que dans la robustesse de segmentation par rapport à la résolution de partition. L'évaluation comparative de la méthode de partition avec ses concurrentes confirme ses avantages en termes de temps d'exécution et de qualité des partitions produites. Par comparaison avec l'approche appliquée directement sur les voxels de l'image, l'étape de partition améliore à la fois le temps d'exécution global et l'empreinte mémoire du processus de segmentation jusqu'à un ordre de grandeur, sans compromettre la qualité de la segmentation en pratique. / We develop a generic Graph Cut-based semiautomatic multiobject image segmentation method principally for use in routine medical applications ranging from tasks involving few objects in 2D images to fairly complex near whole-body 3D image segmentation. The flexible formulation of the method allows its straightforward adaption to a given application.\linebreak In particular, the graph-based vicinity prior model we propose, defined as shortest-path pairwise constraints on the object adjacency graph, can be easily reformulated to account for the spatial relationships between objects in a given problem instance. The segmentation algorithm can be tailored to the runtime requirements of the application and the online storage capacities of the computing platform by an efficient and controllable Voronoi tessellation clustering of the input image which achieves a good balance between cluster compactness and boundary adherence criteria. Qualitative and quantitative comprehensive evaluation and comparison with the standard Potts model confirm that the vicinity prior model brings significant improvements in the correct segmentation of distinct objects of identical intensity, the accurate placement of object boundaries and the robustness of segmentation with respect to clustering resolution. Comparative evaluation of the clustering method with competing ones confirms its benefits in terms of runtime and quality of produced partitions. Importantly, compared to voxel segmentation, the clustering step improves both overall runtime and memory footprint of the segmentation process up to an order of magnitude virtually without compromising the segmentation quality.

Segmentar uma imagem consiste em particioná-la em regiões relevantes para uma dada aplicação, como para isolar um objeto de interesse no domínio de uma imagem. A segmentação é um dos problemas mais fundamentais e desafiadores em processamento de imagem e visão computacional. Ela tem desempenhado um papel importante, por exemplo, na pesquisa em neurologia, envolvendo imagens de Ressonância Magnética (RM), para fins de diagnóstico e tratamento de doenças relacionadas com alterações na anatomia do cérebro humano. Métodos de segmentação baseados na transformada imagem- floresta (IFT, Image Foresting Transform), com funções de conexidade suaves, possuem resultados ótimos, segundo o critério da otimalidade dos caminhos descrito no artigo original da IFT, e têm sido usados com sucesso em várias aplicações, como por exemplo na segmentação de imagens RM de 1.5 Tesla. No entanto, esses métodos carecem de restrições de regularização de borda, podendo gerar segmentações com fronteiras muito irregulares e indesejadas. Eles também não distinguem bem entre bordas similares com orientações opostas, e possuem alta sensibilidade à estimativa dos pesos das arestas do grafo, gerando problemas em imagens com efeitos de inomogeneidade. Nesse trabalho são propostas extensões da IFT, do ponto de vista teórico e experimental, através do uso de funções de conexidade não suaves, para a segmentação interativa de imagens por região. A otimalidade dos novos métodos é suportada pela maximização de energias de corte em grafo, ou como o fruto de uma sequência de iterações de otimização de caminhos em grafos residuais. Como resultados principais temos: O projeto de funções de conexidade mais adaptativas e flexíveis, com o uso de pesos dinâmicos, que permitem um melhor tratamento de imagens com forte inomogeneidade. O uso de grafos direcionados, de modo a explorar a polaridade de borda dos objetos na segmentação por região, e o uso de restrições de forma que ajudam a regularizar a fronteira delineada, favorecendo a segmentação de objetos com formas mais regulares. Esses avanços só foram possíveis devido ao uso de funções não suaves. Portanto, a principal contribuição desse trabalho consiste no suporte teórico para o uso de funções não suaves, até então evitadas na literatura, abrindo novas perpectivas na pesquisa de processamento de imagens usando grafos. / Segmenting an image consist in to partition it into relevant regions for a given application, as to isolate an object of interest in the domain of an image. Segmentation is one of the most fundamental and challenging problems in image processing and computer vision. It has played an important role, for example, in neurology research, involving images of Magnetic Resonance (MR), for the purposes of diagnosis and treatment of diseases related to changes in the anatomy of the human brain. Segmentation methods based on the Image Foresting Transform (IFT), with smooth connectivity functions, have optimum results, according to the criterion of path optimality described in the original IFT paper, and have been successfully used in many applications as, for example, the segmentation of MR images of 1.5 Tesla. However, these methods present a lack of boundary regularization constraints and may produce segmentations with quite irregular and undesired boundaries. They also do not distinguish well between similar boundaries with opposite orientations, and have high sensitivity to the arc-weight estimation of the graph, producing poor results in images with strong inhomogeneity effects. In this work, we propose extensions of the IFT framework, from the theoretical and experimental points of view, through the use of non-smooth connectivity functions for region-based interactive image segmentation. The optimality of the new methods is supported by the maximization of graph cut energies, or as the result of a sequence of paths optimizations in residual graphs. We have as main results: The design of more adaptive and flexible connectivity functions, with the use of dynamic weights, that allow better handling of images with strong inhomogeneity. The use of directed graphs to exploit the boundary polarity of the objects in region-based segmentation, and the use of shape constraints that help to regularize the segmentation boundary, by favoring the segmentation of objects with more regular shapes. These advances were only made possible by the use of non-smooth functions. Therefore, the main contribution of this work is the theoretical support for the usage of non-smooth functions, which were until now avoided in literature, opening new perspectives in the research of image processing using graphs.

This diploma thesis, entitled "Monitoring System for Determination of Motility and Position of Laboratory Animals After Anesthesia", focuses on the design and implementation of contactless detection of the position of a rat or mouse in an enclosure with a transparent cover. The aim of the semester work is to find suitable methods of realization of contactless detection of rat or mouse position and to automatically determine and display average speed or other movement characteristics. The assignment arose from the needs of animal monitoring after curative intervention and also as a necessary utility for future "shading" animal movement (automatic targeting of the scar on the animal's back). The rat, which is located inside our enclosure, is either moving as standard or is dazed after anesthesia. In this work I deal first with search of automatic monitoring systems for detection of animals in the enclosure. Then in the practical part are tested three types of cameras for visual detection of rat position and a script for automatic detection and analysis of rat movement is designed. The system works like a camera eye which in real time is able to find the area of a black box in its field of view and then limit the detection area to the size of this box and then automatically detects the center of gravity and counts. and evaluates the obtained speed with an average calculated with a test of 10 mice - voices on the screen the mouse status in the previous ten seconds. for no stressed animal The rat that is located inside our enclosure is either standard or movable after anesthesia. In this work I first deal with searches of automatic monitoring systems for detecting the position of animals in the enclosure. Then, in the practical part, I test three types of cameras for image detection of rat position. Evaluation software for motion analysis will largely be solved in the follow-up diploma thesis.Project made like monitoring and detecting software based on OpenCV.

Hlavným zámerom tejto diplomovej práce bola automatická segmentácia lézií sklerózy multiplex na snímkoch MRI. V rámci práce boli otestované najnovšie metódy segmentácie s využitím hlbokých neurónových sietí a porovnané prístupy inicializácie váh sietí pomocou preneseného učenia (transfer learning) a samoriadeného učenia (self-supervised learning). Samotný problém automatickej segmentácie lézií sklerózy multiplex je veľmi náročný, a to primárne kvôli vysokej nevyváženosti datasetu (skeny mozgov zvyčajne obsahujú len malé množstvo poškodeného tkaniva). Ďalšou výzvou je manuálna anotácia týchto lézií, nakoľko dvaja rozdielni doktori môžu označiť iné časti mozgu ako poškodené a hodnota Dice Coefficient týchto anotácií je približne 0,86. Možnosť zjednodušenia procesu anotovania lézií automatizáciou by mohlo zlepšiť výpočet množstva lézií, čo by mohlo viesť k zlepšeniu diagnostiky individuálnych pacientov. Našim cieľom bolo navrhnutie dvoch techník využívajúcich transfer learning na predtrénovanie váh, ktoré by neskôr mohli zlepšiť výsledky terajších segmentačných modelov. Teoretická časť opisuje rozdelenie umelej inteligencie, strojového učenia a hlbokých neurónových sietí a ich využitie pri segmentácii obrazu. Následne je popísaná skleróza multiplex, jej typy, symptómy, diagnostika a liečba. Praktická časť začína predspracovaním dát. Najprv boli skeny mozgu upravené na rovnaké rozlíšenie s rovnakou veľkosťou voxelu. Dôvodom tejto úpravy bolo využitie troch odlišných datasetov, v ktorých boli skeny vytvárané rozličnými prístrojmi od rôznych výrobcov. Jeden dataset taktiež obsahoval lebku, a tak bolo nutné jej odstránenie pomocou nástroju FSL pre ponechanie samotného mozgu pacienta. Využívali sme 3D skeny (FLAIR, T1 a T2 modality), ktoré boli postupne rozdelené na individuálne 2D rezy a použité na vstup neurónovej siete s enkodér-dekodér architektúrou. Dataset na trénovanie obsahoval 6720 rezov s rozlíšením 192 x 192 pixelov (po odstránení rezov, ktorých maska neobsahovala žiadnu hodnotu). Využitá loss funkcia bola Combo loss (kombinácia Dice Loss s upravenou Cross-Entropy). Prvá metóda sa zameriavala na využitie predtrénovaných váh z ImageNet datasetu na enkodér U-Net architektúry so zamknutými váhami enkodéra, resp. bez zamknutia a následného porovnania s náhodnou inicializáciou váh. V tomto prípade sme použili len FLAIR modalitu. Transfer learning dokázalo zvýšiť sledovanú metriku z hodnoty približne 0,4 na 0,6. Rozdiel medzi zamknutými a nezamknutými váhami enkodéru sa pohyboval okolo 0,02. Druhá navrhnutá technika používala self-supervised kontext enkodér s Generative Adversarial Networks (GAN) na predtrénovanie váh. Táto sieť využívala všetky tri spomenuté modality aj s prázdnymi rezmi masiek (spolu 23040 obrázkov). Úlohou GAN siete bolo dotvoriť sken mozgu, ktorý bol prekrytý čiernou maskou v tvare šachovnice. Takto naučené váhy boli následne načítané do enkodéru na aplikáciu na náš segmentačný problém. Tento experiment nevykazoval lepšie výsledky, s hodnotou DSC 0,29 a 0,09 (nezamknuté a zamknuté váhy enkodéru). Prudké zníženie metriky mohlo byť spôsobené použitím predtrénovaných váh na vzdialených problémoch (segmentácia a self-supervised kontext enkodér), ako aj zložitosť úlohy kvôli nevyváženému datasetu.

This thesis consider description of segments identified in image. At first there are described main methods of segmentation because it is a process contiguous before describing of objects. Next chapter is devoted to methods which focus on description identified regions. There are studied algorithms used for characterizing of different features. There are parts devoted to color, location, size, orientation, shape and topology. The end of this chapter is devoted to moments. Next chapters are focused on designing fit algorithms for segments description and XML files creating according to MPEG-7 standards and their implementation into RapidMiner. In the last chapter there are described results of the implementation.