Given the importance of color processing in computer vision and computer graphics, estimating and rendering illumination spectral reflectance of image scenes is important to advance the capability of a large class of applications such as scene reconstruction, rendering, surface segmentation, object recognition, and reflectance estimation. Consequently, this dissertation proposes effective methods for reflection components separation and rendering in single scene images. Based on the dichromatic reflectance model, a novel decomposition technique, named the Mean-Shift Decomposition (MSD) method, is introduced to separate the specular from diffuse reflectance components. This technique provides a direct access to surface shape information through diffuse shading pixel isolation. More importantly, this process does not require any local color segmentation process, which differs from the traditional methods that operate by aggregating color information along each image plane. Exploiting the merits of the MSD method, a scene illumination rendering technique is designed to estimate the relative contributing specular reflectance attributes of a scene image. The image feature subset targeted provides a direct access to the surface illumination information, while a newly introduced efficient rendering method reshapes the dynamic range distribution of the specular reflectance components over each image color channel. This image enhancement technique renders the scene illumination reflection effectively without altering the scene’s surface diffuse attributes contributing to realistic rendering effects. As an ancillary contribution, an effective color constancy algorithm based on the dichromatic reflectance model was also developed. This algorithm selects image highlights in order to extract the prominent surface reflectance that reproduces the exact illumination chromaticity. This evaluation is presented using a novel voting scheme technique based on histogram analysis. In each of the three main contributions, empirical evaluations were performed on synthetic and real-world image scenes taken from three different color image datasets. The experimental results show over 90% accuracy in illumination estimation contributing to near real world illumination rendering effects.

[pt] A interpretação de dados sísmicos é de fundamental importância para a industria de óleo e gás. Uma vez que esses tipos de dados possuem um caráter volumétrico, não é tão simples se identificar e selecionar atributos presentes em sua estrutura 3D. Além disso, a grande presença de ruídos e concavidades acentuadas nesse tipo de dado aumenta a complexidade de sua manipulação e visualização. Devido a essas características, a geometria do dado é muito complexa, sendo necessários modelos de iluminação mais realísticos para realizar a iluminação do volume sísmico. Este trabalho consiste em realizar a visualização volumétrica de dados sísmicos baseada no algoritmo de traçado de raios, utilizando um modelo de iluminação por oclusão direcional, calculando a contribuição de luz ambiente que chega a cada elemento do volume. Desta forma, conseguimos realçar a geometria do dado sísmico, sobretudo onde as concavidades e falhas são mais acentuadas. O algoritmo proposto foi inteiramente implementado em placa gráfica, permitindo manipulação a taxas interativas, sem a necessidade de pré-processamento. / [en] The interpretation of seismic volumetric data has a major importance for the oil and gas industry. Since these data types have a volumetric character mode, identify and select attributes present in this struct become a difficult task. Furthemore, the high-frequecy noise and depth information typically found in this type of data, increasesthe complexity of their manipulation and visualization. Due to these characteristics, the geometry of 3D sismic data is very complexy and is necessary more realistic light model to perfom the illumnination of the seismic volume. This work consists of performing a volumetric visualization of seismic data based on ray tracing algorithm, using an illumination model by directional occlusion, computing the ambiente light attenuated by the elements in the light trajetory for all elements in the volume. Thus, we emphasize the geometry of the seismic data, especially the depth cues and spatial relationship. The proposed algorithm was fully implemented on graphics card, allowing at interactive rates, without any pre-processing.

<p>The accessibility and flexibility of mobile devices make them an advantageous platform for gaming, but there are hardware limitations that impede the rendering of high-quality graphics. Rendering complex graphics on a mobile device typically results in a delayed image, also known as latency, and is a great discomfort for users of any real-time rendering experience. This study tests the image stream optimization View Bypass within a cloud gaming architecture, surpassing this imposing limitation by processing the high-quality game render on a remote computational server. A two sample for means test is performed to determine significance between two treatments: the control group without the View Bypass algorithm and the experimental group rendering with the View Bypass algorithm. A SSIM index score is calculated comparing the disparity between the remote server image output and the final mobile device image output after optimizations have been performed. This score indicates the overall image structural integrity difference between the two treatments and determines the quality and effectiveness of the tested algorithm.</p>

The master’s thesis submitted deals with realtime rendering of three-dimensional CAD data using Vulkan API. The thesis also covers possibilities of connecting the API with VRUT, complex solution for renderig developed by ŠKODA AUTO a.s. Design of such connection is presented in form of Vulkan rendering plugin for VRUT application. Performance of the designed module is compared with another rendering module, based on different rendering technologies.

The dissertation deals with utilisation of multispectral optical measurement for data fusion that may be used for visual telepresence and indoor/outdoor mapping by heterogeneous mobile robotic system. Optical proximity sensors, thermal imagers, and tricolour cameras are used for the fusion. The described algorithms are optimised to work in real-time and implemented on CASSANDRA robotic system made by our robotic research group.

This work deals with the 2D and 3D visualization of simulation outputs from the k-Wave toolbox. This toolbox, designed to accurately model the propagation of ultrasound waves in the human body, usually generates immense amounts of output data (up to hundreds of GB). That is why new methods for both the visualization and the effective data representation are necessary to be developed to help users to easily understand the simulation results. This thesis elaborates on the data format, simulation outputs are stored in, with the use of the HDF5 library and looking for the best way to quickly read the simulation data. Finally, the thesis presents the design and the implementation of the console-based application for big simulation data pre-processing and the GUI-based application for interactive visualization of the pre-processed data. The most significant features of these applications are downsampling data, changing the format of storing, viewing 2D sections, planar and volumetric visualization and animation of the simulation process. The proposed implementation allows parts of the simulation domain to be visualised within tens of milliseconds even if the simulation domain comprises GBs of data - This significantly streamlines the work of scientists and clinicians in the field of HIFU.

This work deals with design and implementation of a game based on the XNA platform. It describes the platform and its possibilities of game development for both the PC and the Xbox 360 console. The implemented game is called NanoHeal and it is about the treatment of various health problems. The work also investigates graphic techniques such as non-photorealistic rendering, the depth of field effect and particle systems. These techniques are used to achieve the distinguished look of the game. Key features of the game were evaluated in the online questionnaire.

Les plantes ont longtemps intrigué les scientifiques, qui, avec son importance vitale pour la planète, sa beauté et l'énorme quantité de formes ayant, les rend un sujet attrayant pour la recherche. Un aspect intéressant est la création d'un modèle virtuel capable de simuler de vraies plantes avec un degré élevé de précision. L'objectif de notre étude est les plantes à fleurs, qui jouent un rôle énorme dans notre vie de fins nutritives et médicales à l'embellissement de l'environnement . L'obtention d'un modèle géométrique exacte d'une fleur est très utile, car elle joue un rôle important dans la validation du modèle virtuel. Par ailleurs, la visualisation de paramètres non directement traçables dans les plantes à fleurs vivantes est d'une grande aide à l'étude de la physiologie. L'énorme biodiversité entre les différentes parties d'un spécimen et entre les différents spécimens fournit une vaste zone d'objectifs qui la synthèse d'image doit contester. Modéliser des fleurs est un sous-ensemble d'un espace de recherche beaucoup plus vaste que la modélisation de plantes. Les plantes à fleurs ont des caractéristiques structurelles qui les rendent différentes des structures d'arbres, d’arbustes ou de l’herbe. A ce jour, on ne tient pas une grande importance à essayer cette ligne de recherche d'une façon particulière et en général a été classé dans le contexte plus large de la modélisation des plantes. Nous avons choisi d’utiliser le «L-systems» pour la procédure de la modélisation, et comme base pour notre recherche. Il y a différents mécanismes de catégorisation topologie de la plante dans chacune des étapes de sa croissance. Pour construire le plan de la structure d'une plante, avec une courte grammaire, quelques lignes étaient quelque chose qui dès le premier moment a suscité l'intérêt et par la suite évolué en quelques systèmes d'interprétation géométriques pour la modélisation des plantes. Notre objectif est d'étudier les moyens efficaces de décrire la structure des plantes à fleurs en utilisant L-systems. Tout d'abord, nous proposons de représenter les formes des feuilles, pétales, étamines, carpelles, etc. Avec une extension de L-systems - un modèle basé sur trois cartes généralisées dimensions - 3Gmaps L-systèmes, qui peut être appliquée avec succès pour la modélisation des plantes à fleurs. La description de la grammaire de la structure des plantes à fleurs fournit un nombre illimité de ses interprétations géométriques. Deuxièmement, nous allons améliorer le processus d'écriture de la grammaire par l'ajout d'une nouvelle fonctionnalité de paramétrage interactif. Troisièmement, nous allons proposer une nouvelle méthode de modélisation inverse des plantes à fleurs, où l'utilisateur peut définir de manière interactive les caractéristiques des fleurs. L'algorithme utilise cette information comme une entrée, qui est ensuite analysée et codée en tant que L -systèmes grammaire. Enfin, nous allons présenter une méthode pour créer des clairières de fleurs virtuelles à l'aide de gestes Kinect. Nous voulons faire remarquer que notre travail a été fait avec la plateforme de logiciel 3Gmaps L- système développé dans le cadre de la thèse d'intégrer toutes les techniques proposées. / Plants have always intrigued scientists as besides of its sheer importance for the earth, their beauty and enormous variety of shapes tempt to thoroughly inquire about its nature. One of the aspects of this inquiry is the creation of the virtual model in order to mimic real plants to a high degree of accuracy. The focus of our study is the flowering plants, which play a huge role in our life from nutritive and medical purposes to beautifying the environment. Obtaining an accurate geometrical model of a flower is quite useful as it plays an important role in the validation of the virtual model. Besides, the visualization of parameters not traceable directly in living flowering plants is a stand-by in studying their physiology. A huge biological diversity both within and between individuals provides a vast area of objectives which the image synthesis must challenge.Flower modelling constitutes a part of a larger research area, plant modelling. Flowering plants have their particular structural features which are different from the structure of trees, bushes or grass. Still not a lot of emphasis has been placed to date on this problem, as it was categorized within the modelling of plants in general. We chose a procedural modeling using L-systems as a base of our research. L-system is a very powerful method of plant simulation. It provides a means of characterizing the topology of a plant at every stage of its growth. Grasping the plant structure with just several lines of grammar attracted immediate interest and later on evolved into several powerful geometrical interpretation system used in plant modelling. Our purpose is to study efficient ways of describing the structure of flowering plants by means of L-systems. First, we will propose to represent the shapes of leafs, petals, stamens, carpels, etc. with an extension of L-systems – a model based on three dimensional generalized maps – 3Gmaps L-systems, which can be successfully applied for the modelling of flowering plants. The grammar description of the structure of the flowering plants provides an unlimited number of its geometrical interpretations. Second, we will improve the process of grammar writing by adding a new functionality of interactive parameter adjustment. Third, we will propose a new method of inverse modelling of flowering plants, where the user can interactively define the flower characteristics. The algorithm uses this information as an input, which is then analyzed and coded as L-systems grammar. Finally, we will present a method for creating virtual glades of flowers using Kinect gestures. We want to remark that our work has been done with 3Gmaps L-system software platform developed in the scope of the thesis to integrate all the proposed techniques.

Segmenting image into multiple meaningful regions is an essential task in Computer Vision. Deep Learning has been highly successful for segmentation, benefiting from the availability of the annotated datasets and deep neural network architectures. However, depth-based hand segmentation, an important application area of semantic segmentation, has yet to benefit from rich and large datasets. In addition, while deep methods provide robust solutions, they are often not efficient enough for low-powered devices. In this thesis, we focus on these two problems. To tackle the problem of lack of rich data, we propose an automatic method for generating high-quality annotations and introduce a large scale hand segmentation dataset. By exploiting the visual cues given by an RGBD sensor and a pair of colored gloves, we automatically generate dense annotations for two-hand segmentation. Our automatic annotation method lowers the cost/complexity of creating high-quality datasets and makes it easy to expand the dataset in the future. To reduce the computational requirement and allow real-time segmentation on low power devices, we propose a new representation and architecture for deep networks that predict segmentation maps based on Voronoi Diagrams. Voronoi Diagrams split space into discrete regions based on proximity to a set of points making them a powerful representation of regions, which we can then use to represent our segmentation outcomes. Specifically, we propose to estimate the location and class for these sets of points, which are then rasterized into an image. Notably, we use a differentiable definition of the Voronoi Diagram based on the softmax operator, enabling its use as a decoder layer in an end-to-end trainable network. As rasterization can take place at any given resolution, our method especially excels at rendering high-resolution segmentation maps, given a low-resolution image. We believe that our new HandSeg dataset will open new frontiers in Hand Segmentation research, and our cost-effective automatic annotation pipeline can benefit other relevant labeling tasks. Our newly proposed segmentation network enables high-quality segmentation representations that are not practically possible on low power devices using existing approaches. / Graduate

Etre capable de capturer et de synthétiser des comportements réalistes est un sujet de recherche important en Informatique Graphique. Mais, pour produire des images réalistes, il ne suffit pas de représenter les matériaux de manière fiable, il faut également prendre en compte leurs évolutions dans le temps. La simulation des effets de vieillissement est le domaine de la recherche qui s’intéresse à la prise en compte de la détérioration des modèles virtuels. Après avoir étudié les travaux précédents dans ce domaine, nous avons constaté qu’il y a un manque d’estimation pour certains paramètres environnementaux importants ainsi qu’un besoin de modèles globaux qui tiennent compte des interactions des facteurs intervenants dans les grandes scènes. De plus, nous avons observées que l’amélioration des techniques actuelles nécessite intégrer les caractéristiques physiques réelles pour éviter la déconnection des processus réels, ce qui doit donc être analysés et compris. Dans cette thèse, nous visons à améliorer la simulation du vieillissement en Informatique Graphique sous les trois angles suivants. Concernant l’estimation des facteurs de vieillissement, nous proposons d’abord de traiter un facteur sous-estimé dans la littérature mais qui est très important dans les processus de vieillissement : le soleil. Plus spécifiquement, nous proposons une technique interactive qui estime l’exposition solaire sur des scènes urbaines détaillées. Compte tenu d’une carte d’exposition directionnelle calculée sur une période donnée, nous estimons le facteur de visibilité du ciel qui sert à évaluer l’exposition finale à chaque point visible. Ceci est fait en utilisant une méthode espace-écran basée sur une approche à deux échelles, qui est indépendante de la géométrie et qui a un coût bas en stockage. Notre méthode aboutit à une technique intéressante non seulement pour les simulations de vieillissement, mais également en architecture, dans la conception de bâtiments durables et dans l’estimation de la performance énergétique des bâtiments. Ensuite, nous introduisons un modèle global permettant de simuler les changements d’apparence des bâtiments liés aux dépôts de pollution. Ce processus est très visible sur les scènes urbaines et, jusqu’à présent, aucune technique n’a été proposée en utilisant une approche globale (sur une ville entière). Nous proposons ici une technique basée sur la physique considérant que les effets de la pollution dépendent de trois facteurs principaux : l’exposition au vent, à la pluie et au soleil. Cette approche prend en compte trois étapes : le dépôt, la réaction et le lavage. Avec un pré-calcul à bas coût, nous évaluons la distribution de la pollution dans la ville. Puis, par le biais d’opérateurs espace-écran, nous proposons une méthode efficace capable de générer des images réalistes en temps interactifs. De plus, le pré-calcul n’exige qu’une quantité réduite de mémoire pour stocker la carte de pollution et, comme la méthode utilisée est indépendante de la complexité de la scène, elle peut s’adapter à des modèles complexes et de grande taille en ajustant la résolution de la carte. Finalement, afin de mieux appréhender les processus de vieillissement, nous présentons une méthode permettant de prendre en compte l’effet d’écaillage sur les monuments en pierre avec très peu d’interaction de l’utilisateur. / Being able to capture and synthesize realistic materials is an important research topic in Computer Graphics. In order to render realistic images it is not enough to represent materials reliably, but we also need to represent its temporal component. Simulation of weathering effects is the research area that pursues the incorporation of decay to virtual models. After studying the previous work, we have found that there is a lack of estimation methods for some important environmental parameters and a need of global models that consider the interactions of the environmental factors in big scenes. Moreover, we have observed that the improvement of current techniques need to incorporate real effects characteristics to avoid the current disconnection from the real processes. But, this need to be first analyzed and understood. In this thesis, we aim to improve weathering simulation in Computer Graphics in these three fronts. Concerning weathering factors estimation, we first propose to deal with an undervalued factor in weathering literature which is very important in aging processes: the sun. More specifically, we propose an interactive technique that estimates solar exposure on detailed urban scenes. Given a directional exposure map computed over a given time period, we estimate the sky visibility factor that serves to evaluate the final exposure at each visible point. This is done using a screen-space method based on a two-scale approach, which is geometry independent and has low storage costs. Our method results in an interesting technique not only in weathering simulations, but also in architecture, in sustainable building design and in the estimation of buildings’ energy performance. After that, we introduce a global model to simulate the changes of appearance of buildings due to pollution deposition. This process is very noticeable on urban scenes and, so far, no technique has dealt with it using a global approach. We propose a physically-based technique that considers pollution effects to depend on three main factors: wind, rain and sun exposure, and that takes into account three intervening steps: deposition, reaction and washing. With a low-cost pre-computation step, we evaluate the pollution distribution through the city. Based on this distribution and the use of screen-space operators, our method results in an efficient approach that generate realistic images of polluted surfaces at interactive rates. In addition, the pre-computation demands a reduced amount of memory to store the pollution map and, as it is independent from scene complexity, it can suit large and complex models adapting its resolution. Finally, in order to improve the understanding of weathering processes, we present a method to identify and map “scaling” weathering effects on stone monuments with very little user interaction.