L'imagerie numérique, de la synthèse d'images à la vision par ordinateur est en train de connaître une forte évolution, due entre autres facteurs à la démocratisation et au succès commercial des caméras 3D. Dans le même contexte, l'impression 3D grand public, qui est en train de vivre un essor fulgurant, contribue à la forte demande sur ce type de caméra pour les besoins de la numérisation 3D. L'objectif de cette thèse est d'acquérir et de maîtriser un savoir-faire dans le domaine de la capture/acquisition de modèles 3D en particulier sur l'aspect rendu réaliste. La réalisation d'un scanner 3D à partir d'une caméra RGB-D fait partie de l'objectif. Lors de la phase d'acquisition, en particulier pour un dispositif portable, on est confronté à deux problèmes principaux, le problème lié au référentiel de chaque capture et le rendu final de l'objet reconstruit. / Digital imaging, from the synthesis of images to computer vision isexperiencing a strong evolution, due among other factors to the democratization and commercial success of 3D cameras. In the same context, the consumer 3D printing, which is experiencing a rapid rise, contributes to the strong demand for this type of camera for the needs of 3D scanning. The objective of this thesis is to acquire and master a know-how in the field of the capture / acquisition of 3D models in particular on the rendered aspect. The realization of a 3D scanner from a RGB-D camera is part of the goal. During the acquisition phase, especially for a portable device, there are two main problems, the problem related to the repository of each capture and the final rendering of the reconstructed object.

L'accès aux séquences d'images 3D s'est aujourd'hui démocratisé, grâce aux récentes avancées dans le développement des capteurs de profondeur ainsi que des méthodes permettant de manipuler des informations 3D à partir d'images 2D. De ce fait, il y a une attente importante de la part de la communauté scientifique de la vision par ordinateur dans l'intégration de l'information 3D. En effet, des travaux de recherche ont montré que les performances de certaines applications pouvaient être améliorées en intégrant l'information 3D. Cependant, il reste des problèmes à résoudre pour l'analyse et la segmentation de scènes intérieures comme (a) comment l'information 3D peut-elle être exploitée au mieux ? et (b) quelle est la meilleure manière de prendre en compte de manière conjointe les informations couleur et 3D ? Nous abordons ces deux questions dans cette thèse et nous proposons de nouvelles méthodes non supervisées pour la classification d'images 3D et la segmentation prenant en compte de manière conjointe les informations de couleur et de profondeur. A cet effet, nous formulons l'hypothèse que les normales aux surfaces dans les images 3D sont des éléments à prendre en compte pour leur analyse, et leurs distributions sont modélisables à l'aide de lois de mélange. Nous utilisons la méthode dite « Bregman Soft Clustering » afin d'être efficace d'un point de vue calculatoire. De plus, nous étudions plusieurs lois de probabilités permettant de modéliser les distributions de directions : la loi de von Mises-Fisher et la loi de Watson. Les méthodes de classification « basées modèles » proposées sont ensuite validées en utilisant des données de synthèse puis nous montrons leur intérêt pour l'analyse des images 3D (ou de profondeur). Une nouvelle méthode de segmentation d'images couleur et profondeur, appelées aussi images RGB-D, exploitant conjointement la couleur, la position 3D, et la normale locale est alors développée par extension des précédentes méthodes et en introduisant une méthode statistique de fusion de régions « planes » à l'aide d'un graphe. Les résultats montrent que la méthode proposée donne des résultats au moins comparables aux méthodes de l'état de l'art tout en demandant moins de temps de calcul. De plus, elle ouvre des perspectives nouvelles pour la fusion non supervisée des informations de couleur et de géométrie. Nous sommes convaincus que les méthodes proposées dans cette thèse pourront être utilisées pour la classification d'autres types de données comme la parole, les données d'expression en génétique, etc. Elles devraient aussi permettre la réalisation de tâches complexes comme l'analyse conjointe de données contenant des images et de la parole / Access to the 3D images at a reasonable frame rate is widespread now, thanks to the recent advances in low cost depth sensors as well as the efficient methods to compute 3D from 2D images. As a consequence, it is highly demanding to enhance the capability of existing computer vision applications by incorporating 3D information. Indeed, it has been demonstrated in numerous researches that the accuracy of different tasks increases by including 3D information as an additional feature. However, for the task of indoor scene analysis and segmentation, it remains several important issues, such as: (a) how the 3D information itself can be exploited? and (b) what is the best way to fuse color and 3D in an unsupervised manner? In this thesis, we address these issues and propose novel unsupervised methods for 3D image clustering and joint color and depth image segmentation. To this aim, we consider image normals as the prominent feature from 3D image and cluster them with methods based on finite statistical mixture models. We consider Bregman Soft Clustering method to ensure computationally efficient clustering. Moreover, we exploit several probability distributions from directional statistics, such as the von Mises-Fisher distribution and the Watson distribution. By combining these, we propose novel Model Based Clustering methods. We empirically validate these methods using synthetic data and then demonstrate their application for 3D/depth image analysis. Afterward, we extend these methods to segment synchronized 3D and color image, also called RGB-D image. To this aim, first we propose a statistical image generation model for RGB-D image. Then, we propose novel RGB-D segmentation method using a joint color-spatial-axial clustering and a statistical planar region merging method. Results show that, the proposed method is comparable with the state of the art methods and requires less computation time. Moreover, it opens interesting perspectives to fuse color and geometry in an unsupervised manner. We believe that the methods proposed in this thesis are equally applicable and extendable for clustering different types of data, such as speech, gene expressions, etc. Moreover, they can be used for complex tasks, such as joint image-speech data analysis

The Professorship of Digital Signal Processing and Circuit Technology of Chemnitz University of Technology conducts research in the field of three-dimensional space measurement with optical sensors. In recent years this field has made major progress. For example innovative, active techniques such as the “structured light“-principle are able to measure even homogeneous surfaces and find its way into the consumer electronic market in terms of Microsoft’s Kinect® at the present time. Furthermore, high-resolution optical sensors establish powerful, passive stereo vision systems in the field of indoor surveillance. Thereby they induce new application domains such as security and assistance systems for domestic environments. However, the constraint field of view can be still considered as an essential characteristic of all these technologies. For instance, in order to measure a volume in size of a living space, two to three deployed 3D sensors have to be applied nowadays. This is due to the fact that the commonly utilized perspective projection principle constrains the visible area to a field of view of approximately 120°. On the contrary, novel fish-eye lenses allow the realization of omnidirectional projection models. Therewith, the visible field of view can be enlarged up to more than 180°. In combination with a 3D measurement approach, thus, the number of required sensors for entire room coverage can be reduced considerably. Motivated by the requirements of the field of indoor surveillance, the present work focuses on the combination of the established stereo vision principle and omnidirectional projection methods. The entire 3D measurement of a living space by means of one single sensor can be considered as major objective. As a starting point for this thesis chapter 1 discusses the underlying requirement, referring to various relevant fields of application. Based on this, the distinct purpose for the present work is stated. The necessary mathematical foundations of computer vision are reflected in Chapter 2 subsequently. Based on the geometry of the optical imaging process, the projection characteristics of relevant principles are discussed and a generic method for modeling fish-eye cameras is selected. Chapter 3 deals with the extraction of depth information using classical (perceptively imaging) binocular stereo vision configurations. In addition to a complete recap of the processing chain, especially occurring measurement uncertainties are investigated. In the following, Chapter 4 addresses special methods to convert different projection models. The example of mapping an omnidirectional to a perspective projection is employed, in order to develop a method for accelerating this process and, hereby, for reducing the computational load associated therewith. Any errors that occur, as well as the necessary adjustment of image resolution, are an integral part of the investigation. As a practical example, an application for person tracking is utilized in order to demonstrate to which extend the usage of “virtual views“ can increase the recognition rate for people detectors in the context of omnidirectional monitoring. Subsequently, an extensive search with respect to omnidirectional imaging stereo vision techniques is conducted in chapter 5. It turns out that the complete 3D capture of a room is achievable by the generation of a hemispherical depth map. Therefore, three cameras have to be combined in order to form a trinocular stereo vision system. As a basis for further research, a known trinocular stereo vision method is selected. Furthermore, it is hypothesized that, applying a modified geometric constellation of cameras, more precisely in the form of an equilateral triangle, and using an alternative method to determine the depth map, the performance can be increased considerably. A novel method is presented, which shall require fewer operations to calculate the distance information and which is to avoid a computational costly step for depth map fusion as necessary in the comparative method. In order to evaluate the presented approach as well as the hypotheses, a hemispherical depth map is generated in Chapter 6 by means of the new method. Simulation results, based on artificially generated 3D space information and realistic system parameters, are presented and subjected to a subsequent error estimate. A demonstrator for generating real measurement information is introduced in Chapter 7. In addition, the methods that are applied for calibrating the system intrinsically as well as extrinsically are explained. It turns out that the calibration procedure utilized cannot estimate the extrinsic parameters sufficiently. Initial measurements present a hemispherical depth map and thus con.rm the operativeness of the concept, but also identify the drawbacks of the calibration used. The current implementation of the algorithm shows almost real-time behaviour. Finally, Chapter 8 summarizes the results obtained along the studies and discusses them in the context of comparable binocular and trinocular stereo vision approaches. For example the results of the simulations carried out produced a saving of up to 30% in terms of stereo correspondence operations in comparison with a referred trinocular method. Furthermore, the concept introduced allows the avoidance of a weighted averaging step for depth map fusion based on precision values that have to be calculated costly. The achievable accuracy is still comparable for both trinocular approaches. In summary, it can be stated that, in the context of the present thesis, a measurement system has been developed, which has great potential for future application fields in industry, security in public spaces as well as home environments.

Nowadays, new computers generation provides a high performance that enables to build computationally expensive computer vision applications applied to mobile robotics. Building a map of the environment is a common task of a robot and is an essential part to allow the robots to move through these environments. Traditionally, mobile robots used a combination of several sensors from different technologies. Lasers, sonars and contact sensors have been typically used in any mobile robotic architecture, however color cameras are an important sensor due to we want the robots to use the same information that humans to sense and move through the different environments. Color cameras are cheap and flexible but a lot of work need to be done to give robots enough visual understanding of the scenes. Computer vision algorithms are computational complex problems but nowadays robots have access to different and powerful architectures that can be used for mobile robotics purposes. The advent of low-cost RGB-D sensors like Microsoft Kinect which provide 3D colored point clouds at high frame rates made the computer vision even more relevant in the mobile robotics field. The combination of visual and 3D data allows the systems to use both computer vision and 3D processing and therefore to be aware of more details of the surrounding environment. The research described in this thesis was motivated by the need of scene mapping. Being aware of the surrounding environment is a key feature in many mobile robotics applications from simple robotic navigation to complex surveillance applications. In addition, the acquisition of a 3D model of the scenes is useful in many areas as video games scene modeling where well-known places are reconstructed and added to game systems or advertising where once you get the 3D model of one room the system can add furniture pieces using augmented reality techniques. In this thesis we perform an experimental study of the state-of-the-art registration methods to find which one fits better to our scene mapping purposes. Different methods are tested and analyzed on different scene distributions of visual and geometry appearance. In addition, this thesis proposes two methods for 3d data compression and representation of 3D maps. Our 3D representation proposal is based on the use of Growing Neural Gas (GNG) method. This Self-Organizing Maps (SOMs) has been successfully used for clustering, pattern recognition and topology representation of various kind of data. Until now, Self-Organizing Maps have been primarily computed offline and their application in 3D data has mainly focused on free noise models without considering time constraints. Self-organising neural models have the ability to provide a good representation of the input space. In particular, the Growing Neural Gas (GNG) is a suitable model because of its flexibility, rapid adaptation and excellent quality of representation. However, this type of learning is time consuming, specially for high-dimensional input data. Since real applications often work under time constraints, it is necessary to adapt the learning process in order to complete it in a predefined time. This thesis proposes a hardware implementation leveraging the computing power of modern GPUs which takes advantage of a new paradigm coined as General-Purpose Computing on Graphics Processing Units (GPGPU). Our proposed geometrical 3D compression method seeks to reduce the 3D information using plane detection as basic structure to compress the data. This is due to our target environments are man-made and therefore there are a lot of points that belong to a plane surface. Our proposed method is able to get good compression results in those man-made scenarios. The detected and compressed planes can be also used in other applications as surface reconstruction or plane-based registration algorithms. Finally, we have also demonstrated the goodness of the GPU technologies getting a high performance implementation of a CAD/CAM common technique called Virtual Digitizing.

This thesis presents an approach for configuring and calibrating a network of RGB-D sensors used to guide a robotic arm to interact with objects that get rapidly modeled in 3D. The system is based on Microsoft Kinect sensors for 3D data acquisition. The work presented here also details an analysis and experimental study of the Kinect’s depth sensor capabilities and performance. The study comprises examination of the resolution, quantization error, and random distribution of depth data. In addition, the effects of color and reflectance characteristics of an object are also analyzed. The study examines two versions of Kinect sensors, one dedicated to operate with the Xbox 360 video game console and the more recent Microsoft Kinect for Windows version. The study of the Kinect sensor is extended to the design of a rapid acquisition system dedicated to large workspaces by the linkage of multiple Kinect units to collect 3D data over a large object, such as an automotive vehicle. A customized calibration method for this large workspace is proposed which takes advantage of the rapid 3D measurement technology embedded in the Kinect sensor and provides registration accuracy between local sections of point clouds that is within the range of the depth measurements accuracy permitted by the Kinect technology. The method is developed to calibrate all Kinect units with respect to a reference Kinect. The internal calibration of the sensor in between the color and depth measurements is also performed to optimize the alignment between the modalities. The calibration of the 3D vision system is also extended to formally estimate its configuration with respect to the base of a manipulator robot, therefore allowing for seamless integration between the proposed vision platform and the kinematic control of the robot. The resulting vision-robotic system defines the comprehensive calibration of reference Kinect with the robot. The latter can then be used to interact under visual guidance with large objects, such as vehicles, that are positioned within a significantly enlarged field of view created by the network of RGB-D sensors. The proposed design and calibration method is validated in a real world scenario where five Kinect sensors operate collaboratively to rapidly and accurately reconstruct a 180 degrees coverage of the surface shape of various types of vehicles from a set of individual acquisitions performed in a semi-controlled environment, that is an underground parking garage. The vehicle geometrical properties generated from the acquired 3D data are compared with the original dimensions of the vehicle.

This thesis presents an approach for configuring and calibrating a network of RGB-D sensors used to guide a robotic arm to interact with objects that get rapidly modeled in 3D. The system is based on Microsoft Kinect sensors for 3D data acquisition. The work presented here also details an analysis and experimental study of the Kinect’s depth sensor capabilities and performance. The study comprises examination of the resolution, quantization error, and random distribution of depth data. In addition, the effects of color and reflectance characteristics of an object are also analyzed. The study examines two versions of Kinect sensors, one dedicated to operate with the Xbox 360 video game console and the more recent Microsoft Kinect for Windows version. The study of the Kinect sensor is extended to the design of a rapid acquisition system dedicated to large workspaces by the linkage of multiple Kinect units to collect 3D data over a large object, such as an automotive vehicle. A customized calibration method for this large workspace is proposed which takes advantage of the rapid 3D measurement technology embedded in the Kinect sensor and provides registration accuracy between local sections of point clouds that is within the range of the depth measurements accuracy permitted by the Kinect technology. The method is developed to calibrate all Kinect units with respect to a reference Kinect. The internal calibration of the sensor in between the color and depth measurements is also performed to optimize the alignment between the modalities. The calibration of the 3D vision system is also extended to formally estimate its configuration with respect to the base of a manipulator robot, therefore allowing for seamless integration between the proposed vision platform and the kinematic control of the robot. The resulting vision-robotic system defines the comprehensive calibration of reference Kinect with the robot. The latter can then be used to interact under visual guidance with large objects, such as vehicles, that are positioned within a significantly enlarged field of view created by the network of RGB-D sensors. The proposed design and calibration method is validated in a real world scenario where five Kinect sensors operate collaboratively to rapidly and accurately reconstruct a 180 degrees coverage of the surface shape of various types of vehicles from a set of individual acquisitions performed in a semi-controlled environment, that is an underground parking garage. The vehicle geometrical properties generated from the acquired 3D data are compared with the original dimensions of the vehicle.

Dans cette thèse, nous traitons des problèmes de la détection des personnes et de l'estimation de leurs poses dans la Salle Opératoire (SO), deux éléments clés pour le développement d'applications d'assistance chirurgicale. Nous percevons la salle grâce à des caméras RGB-D qui fournissent des informations visuelles complémentaires sur la scène. Ces informations permettent de développer des méthodes mieux adaptées aux difficultés propres aux SO, comme l'encombrement, les surfaces sans texture et les occlusions. Nous présentons des nouvelles approches qui tirent profit des informations temporelles, de profondeur et des vues multiples afin de construire des modèles robustes pour la détection des personnes et de leurs poses. Une évaluation est effectuée sur plusieurs jeux de données complexes enregistrés dans des salles opératoires avec une ou plusieurs caméras. Les résultats obtenus sont très prometteurs et montrent que nos approches surpassent les méthodes de l'état de l'art sur ces données cliniques. / In this thesis, we address the two problems of person detection and pose estimation in Operating Rooms (ORs), which are key ingredients in the development of surgical assistance applications. We perceive the OR using compact RGB-D cameras that can be conveniently integrated in the room. These sensors provide complementary information about the scene, which enables us to develop methods that can cope with numerous challenges present in the OR, e.g. clutter, textureless surfaces and occlusions. We present novel part-based approaches that take advantage of depth, multi-view and temporal information to construct robust human detection and pose estimation models. Evaluation is performed on new single- and multi-view datasets recorded in operating rooms. We demonstrate very promising results and show that our approaches outperform state-of-the-art methods on this challenging data acquired during real surgeries.

Robots are operating for longer times and collecting much more data than just a few years ago. In this setting we are interested in exploring ways of modeling the environment, segmenting out areas of interest and keeping track of the segmentations over time, with the purpose of building 4D models (i.e. space and time) of the relevant parts of the environment. Our approach relies on repeatedly observing the environment and creating local maps at specific locations. The first question we address is how to choose where to build these local maps. Traditionally, an operator defines a set of waypoints on a pre-built map of the environment which the robot visits autonomously. Instead, we propose a method to automatically extract semantically meaningful regions from a point cloud representation of the environment. The resulting segmentation is purely geometric, and in the context of mobile robots operating in human environments, the semantic label associated with each segment (i.e. kitchen, office) can be of interest for a variety of applications. We therefore also look at how to obtain per-pixel semantic labels given the geometric segmentation, by fusing probabilistic distributions over scene and object types in a Conditional Random Field. For most robotic systems, the elements of interest in the environment are the ones which exhibit some dynamic properties (such as people, chairs, cups, etc.), and the ability to detect and segment such elements provides a very useful initial segmentation of the scene. We propose a method to iteratively build a static map from observations of the same scene acquired at different points in time. Dynamic elements are obtained by computing the difference between the static map and new observations. We address the problem of clustering together dynamic elements which correspond to the same physical object, observed at different points in time and in significantly different circumstances. To address some of the inherent limitations in the sensors used, we autonomously plan, navigate around and obtain additional views of the segmented dynamic elements. We look at methods of fusing the additional data and we show that both a combined point cloud model and a fused mesh representation can be used to more robustly recognize the dynamic object in future observations. In the case of the mesh representation, we also show how a Convolutional Neural Network can be trained for recognition by using mesh renderings. Finally, we present a number of methods to analyse the data acquired by the mobile robot autonomously and over extended time periods. First, we look at how the dynamic segmentations can be used to derive a probabilistic prior which can be used in the mapping process to further improve and reinforce the segmentation accuracy. We also investigate how to leverage spatial-temporal constraints in order to cluster dynamic elements observed at different points in time and under different circumstances. We show that by making a few simple assumptions we can increase the clustering accuracy even when the object appearance varies significantly between observations. The result of the clustering is a spatial-temporal footprint of the dynamic object, defining an area where the object is likely to be observed spatially as well as a set of time stamps corresponding to when the object was previously observed. Using this data, predictive models can be created and used to infer future times when the object is more likely to be observed. In an object search scenario, this model can be used to decrease the search time when looking for specific objects. / <p>QC 20171009</p>

This thesis presents an approach for configuring and calibrating a network of RGB-D sensors used to guide a robotic arm to interact with objects that get rapidly modeled in 3D. The system is based on Microsoft Kinect sensors for 3D data acquisition. The work presented here also details an analysis and experimental study of the Kinect’s depth sensor capabilities and performance. The study comprises examination of the resolution, quantization error, and random distribution of depth data. In addition, the effects of color and reflectance characteristics of an object are also analyzed. The study examines two versions of Kinect sensors, one dedicated to operate with the Xbox 360 video game console and the more recent Microsoft Kinect for Windows version. The study of the Kinect sensor is extended to the design of a rapid acquisition system dedicated to large workspaces by the linkage of multiple Kinect units to collect 3D data over a large object, such as an automotive vehicle. A customized calibration method for this large workspace is proposed which takes advantage of the rapid 3D measurement technology embedded in the Kinect sensor and provides registration accuracy between local sections of point clouds that is within the range of the depth measurements accuracy permitted by the Kinect technology. The method is developed to calibrate all Kinect units with respect to a reference Kinect. The internal calibration of the sensor in between the color and depth measurements is also performed to optimize the alignment between the modalities. The calibration of the 3D vision system is also extended to formally estimate its configuration with respect to the base of a manipulator robot, therefore allowing for seamless integration between the proposed vision platform and the kinematic control of the robot. The resulting vision-robotic system defines the comprehensive calibration of reference Kinect with the robot. The latter can then be used to interact under visual guidance with large objects, such as vehicles, that are positioned within a significantly enlarged field of view created by the network of RGB-D sensors. The proposed design and calibration method is validated in a real world scenario where five Kinect sensors operate collaboratively to rapidly and accurately reconstruct a 180 degrees coverage of the surface shape of various types of vehicles from a set of individual acquisitions performed in a semi-controlled environment, that is an underground parking garage. The vehicle geometrical properties generated from the acquired 3D data are compared with the original dimensions of the vehicle.

With the release of the Kinect device new possibilities appeared, allowing a simple use of image depth in image processing. The aim of this thesis is to propose a method for object detection and recognition in a depth map. Well known method Bag of Words and a descriptor based on Spin Image method are used for the object recognition. The Spin Image method is one of several existing approaches to depth map which are described in this thesis. Detection of object in picture is ensured by the sliding window technique. That is improved and speeded up by utilization of the depth information.