Tesis sobre el tema "Stereo baseline"

Like humans, computer vision systems can better infer a scene's 3-D structure by processing its 2-D images taken from multiple viewpoints. While this seems effortless for humans, it is still a challenge for computer vision. Underlying the act of associating the different perspectives is a problem called wide-baseline stereo, which computes the geometric relationship between two overlapping views. Wide-baseline stereo can be problematic when working on images taken of real-life urban environments, due to practical issues such as poor image quality or ambiguity raised by repetitive patterns. We analyze why these factors pose difficulties for current methods and propose principles that can make wide-baseline stereo more effective, in terms of both robustness and accuracy. We treat wide-baseline stereo as a sequence of three sub-problems: feature detection, feature matching, and fundamental matrix estimation. We propose improvements for each of these and test them on real images of 3-D urban scenes. For feature detection, we demonstrate that when we use both image intensity contrast and entropy-based visual saliency, we are better at repeatably extracting features of a 3-D scene. We use intensity contrast as a cue for obtaining initial feature seeds, which are then evaluated and locally adapted according to an entropy-based saliency measure. We select features with high saliency scores. Experimental comparisons against peer feature detectors show that our method detects more regular structures and fewer noisy patterns. As a result, our method detects features with high repeatability, which is conducive to the subsequent feature matching. In the case of feature matching, we show that we can match features more robustly when using both local feature appearance and regional image information. We model global image information with a graph, whose nodes contain local feature appearances and edges encode semi-local proximity structure. Working on this graph, we convert t
L'utilisation de plusieurs points de vue d'une scène pour en déterminer sa structure tridimensionnelle est un exercice effectué autant par l'humain que par certains systèmes de vision artificielle. Mais alors qu'il ne requiert aucun effort pour l'humain, il représente un défi pour le domaine de la vision par ordinateur. Un problème sous-jacent à celui d'associer plusieurs perspectives d'une scène est celui de la stéréoscopie pour une longue ligne de base, qui consiste à déterminer les relations géométriques entre deux vues qui se chevauchent. La stéréo pour une longue ligne de base (lorsque les deux points de vue sont éloignés) peut être problématique dans un environnement urbain, en raison d'une qualité parfois pauvre des images, et aussi de l'ambiguïté que peut soulever des formes répétitives. Cette thèse analyse les raisons pour lesquelles ces facteurs peuvent être problématiques pour les méthodes actuelles et propose des principes qui permettent une stéréo plus efficace, autant au point de vue de la robustesse que de la précision. La stéréo d'images provenant de points de vues éloignés est divisée en trois sous-problémes: la détection de caractéristiques visuelles, leur appariement, ainsi que l'estimation de la matrice fondamentale. Des améliorations sont proposées pour chaque élément, et des expérimentations sur des données réelles de scènes urbaines sont présentées. Pour la détection de caractéristiques, il est démontré que lorsque le contraste en intensité des images ainsi que la saillance visuelle basée sur l'entropie sont utilisés, nous obtenons de meilleurs résultats de détection de caractéristiques de scènes tridimensionnelles. Le contraste en intensité est utilisé pour obtenir des points de départ pour les caractéristiques, qui sont ensuite évalués et adapté localement selon une mesure de saillance basée sur l'entropie. Les caractéristiques ayant obtenues une mesure élevée de sai

Conventional stereo vision systems rely on two spatially fixed cameras to gather depth information about a scene. The cameras typically have a fixed distance between them, known as the baseline. As the baseline increases, the estimated 3D information becomes more accurate, which makes it advantageous to have as large a baseline as possible. However, large baselines have problems whenever objects approach the cameras. The objects begin to leave the field of view of the cameras, making it impossible to determine where they are located in 3D space. This becomes especially important if an object of interest must be actuated upon and is approached by a vehicle. In an attempt to overcome this limitation, this thesis introduces a variable baseline stereo system that can adjust its baseline automatically based on the location of an object of interest. This allows accurate depth information to be gathered when an object is both near and far. The system was designed to operate under, and automatically move to a large range of different baselines. This thesis presents the mechanical design of the stereo boom. This is followed by a derivation of a control scheme that adjusts the baseline based on an estimate object location, which is gathered from stereo vision. This algorithm ensures that a certain incident angle on an object of interest is never surpassed. This maximum angle is determined by where a stereo correspondence algorithm, Semi-Global Block Matching, fails to create full reconstructions.
Master of Science

This thesis presents online 3D reconstruction and ground segmentation using unmanned aerial vehicle (UAV) based stereo vision. For this purpose, a long baseline stereo vision system has been designed and built. Application of this system is to work as part of an air and ground based multi-robot autonomous terrain surveying project at Unmanned Systems Lab (USL), Virginia Tech, to act as a first responder robotic system in disaster situations. Areas covered by this thesis are design of long baseline stereo vision system, study of stereo vision raw output, techniques to filter out outliers from raw stereo vision output, a 3D reconstruction method and a study to improve running time by controlling the density of point clouds. Presented work makes use of filtering methods and implementations in Point Cloud Library (PCL) and feature matching on graphics processing unit (GPU) using OpenCV with CUDA. Besides 3D reconstruction, the challenge in the project was speed and several steps and ideas are presented to achieve it. Presented 3D reconstruction algorithm uses feature matching in 2D images, converts keypoints to 3D using disparity images, estimates rigid body transformation between matched 3D keypoints and fits point clouds. To correct and control orientation and localization errors, it fits re-projected UAV positions on GPS recorded UAV positions using iterative closest point (ICP) algorithm as the correction step. A new but computationally intensive process of use of superpixel clustering and plane fitting to increase resolution of disparity images to sub-pixel resolution is also presented. Results section provides accuracy of 3D reconstruction results. The presented process is able to generate application acceptable semi-dense 3D reconstruction and ground segmentation at 8-12 frames per second (fps). In 3D reconstruction of an area of size 25 x 40 m2, with UAV flight altitude of 23 m, average obstacle localization error and average obstacle size/dimension error is found to be of 17 cm and 3 cm, respectively.
MS

This thesis addresses the problem of reconstructing the three-dimensional (3D) digital model of a scene from a collection of two-dimensional (2D) images taken from it. To address this fundamental computer vision problem, we propose three algorithms. They are the main contributions of this thesis. First, we solve multiview stereo with the o -axis aperture camera. This system has a very small baseline as images are captured from viewpoints close to each other. The key idea is to change the size or the 3D location of the aperture of the camera so as to extract selected portions of the scene. Our imaging model takes both defocus and stereo information into account and allows to solve shape reconstruction and image restoration in one go. The o -axis aperture camera can be used in a small-scale space where the camera motion is constrained by the surrounding environment, such as in 3D endoscopy. Second, to solve multiview stereo with large baseline, we present a framework that poses the problem of recovering a 3D surface in the scene as a regularized minimal partition problem of a visibility function. The formulation is convex and hence guarantees that the solution converges to the global minimum. Our formulation is robust to view-varying extensive occlusions, clutter and image noise. At any stage during the estimation process the method does not rely on the visual hull, 2D silhouettes, approximate depth maps, or knowing which views are dependent(i.e., overlapping) and which are independent( i.e., non overlapping). Furthermore, the degenerate solution, the null surface, is not included as a global solution in this formulation. One limitation of this algorithm is that its computation complexity grows with the number of views that we combine simultaneously. To address this limitation, we propose a third formulation. In this formulation, the visibility functions are integrated within a narrow band around the estimated surface by setting weights to each point along optical rays. This thesis presents technical descriptions for each algorithm and detailed analyses to show how these algorithms improve existing reconstruction techniques.

Three-dimensional information is often used in robotics and 3D-mapping. There exist several ways to obtain a three-dimensional map. However, the time of flight used in the laser scanners or the structured light utilized by Kinect-like sensors sometimes are not sufficient. In this thesis, we investigate two CNN based stereo matching methods for obtaining 3D-information from a grayscaled pair of rectified images.While the state-of-the-art stereo matching method utilize a Siamese architecture, in this project a two-channel and a two stream network are trained in an attempt to outperform the state-of-the-art. A set of experiments were performed to achieve optimal hyperparameters. By changing one parameter at the time, the networks with architectures mentioned above are trained. After a completed training the networks are evaluated with two criteria, the error rate, and the runtime.Due to time limitations, we were not able to find optimal learning parameters. However, by using settings from [17] we train a two-channel network that performed almost on the same level as the state-of-the-art. The error rate on the test data for our best architecture is 2.64% while the error rate for the state-of-the-art Siamese network is 2.62%. We were not able to achieve better performance than the state-of-the-art, but we believe that it is possible to reduce the error rate further. On the other hand, the state-of-the-art Siamese stereo matching network is more efficient and faster during the disparity estimation. Therefore, if the time efficiency is prioritized, the Siamese based network should be considered.

Depth perception and 3D object recognition with stereo cameras has a variety of applications. For example, the technology enables to sense ob-jects and possible dangers on automated cars using passive and cheap sen-sors. However, sensing range corresponds to camera baseline, and rigid camera mounting is only available to rather small baseline setups. Thus, typical systems do not exceed sensing ranges of around 50m. To enable higher ranges with baselines of about 1-2m on moving and vibrating plat-forms, non-rigidity of the setup (i.e. a variable relative camera orientation) must be considered. Hence, calibration and depth image processing must be extended with re-calibration with every stereo pair along the image se-quence. Ideally, this re-calibration works in real-time and on arbitrary im-age sequences, however this must still be evaluated.

Mestrado em Engenharia Mecânica
Os veículos autónomos são uma tendência cada vez mais crescente nos dias de hoje com os grandes fabricantes da área automóvel, e não só, concentrados em desenvolver carros autónomos. As duas maiores vantagens que se destacam para os carros autónomos são maior conforto para o condutor e maior segurança, onde este trabalho se foca. São incontáveis as vezes que um condutor, por distração ou por outra razão, não vê um objeto na estrada e colide ou um peão na estrada que e atropelado. Esta e uma das questões que um sistema de apoio a condução (ADAS) ou um carro autónomo tenta solucionar e por ser uma questão tão relevante há cada vez mais investigação nesta área. Um dos sistemas mais usados para este tipo de aplicação são câmaras digitais, que fornecem informação muito completa sobre o meio circundante, para além de sistemas como sensores LIDAR, entre outros. Uma tendência que deriva desta e o uso de sistemas stereo, sistemas com duas câmaras, e neste contexto coloca-se uma pergunta a qual este trabalho tenta respoder: "qual e a distância ideal entre as câmaras num sistema stereo para deteção de objetos ou peões?". Esta tese apresenta todo o desenvolvimento de um sistema de visão stereo: desde o desenvolvimento de todo o software necessário para calcular a que distância estão peões e objetos usando duas câmaras até ao desenvolvimento de um sistema de xação das câmaras que permita o estudo da qualidade da deteção de peões para várias baselines. Foram realizadas experiências para estudar a influênci da baseline e da distância focal da lente que consistriam em gravar imagens com um peão em deslocamento a distâncias pré defenidas e marcadas no chão assim como um objeto xo, tudo em cenário exterior. A análise dos resultados foi feita comparando o valor calculado automáticamente pela aplicação com o valor medido. Conclui-se que com este sistema e com esta aplicação e possível detetar peões com exatidão razoável. No entanto, os melhores resultados foram obtidos para a baseline de 0.3m e para uma lente de 8mm.
Nowadays, autonomous vehicles are an increasing trend as the major players of this sector, and not only, are focused in developing autonomous cars. The two main advantages of autonomous cars are the higher convenience for the passengers and more safety for the passengers and for the people around, which is what this thesis focus on. Sometimes, due to distraction or another reasons, the driver does not see an object on the road and crash or a pedestrian in the cross walk and the person is run over. This is one of the questions that an ADAS or an autonomous car tries to solve and due to the huge relevance of this more research have been done in this area. One of the most applied systems for ADAS are digital cameras, that provide complex information about the surrounding environment, in addition to LIDAR sensor and others. Following this trend, the use of stereo vision systems is increasing - systems with two cameras, and in this context a question comes up: "what is the ideal distance between the cameras in a stereo system for object and pedestrian detection?". This thesis shows all the development of a stereo vision system: from the development of the necessary software for calculating the objects and pedestrians distance form the setup using two cameras, to the design of a xing system for the cameras that allows the study of stereo for di erent baselines. In order to study the in uence of the baseline and the focal distance a pedestrian, walking through previously marked positions, and a xed object, were recorded, in an exterior scenario. The results were analyzed by comparing the automatically calculated distance, using the application, with the real value measured. It was concluded, in the end, that the distance of pedestrians and objects can be calculated, with minimal error, using the software developed and the xing support system. However, the best results were achieved for the 0.3m baseline and for the 8mm lens.

Cette thèse a été réalisée dans le secteur de l'industrie automobile, en collaboration avec le Groupe Renault et concerne en particulier le développement de systèmes d'aide à la conduite avancés et de véhicules autonomes. Les progrès réalisés par la communauté scientifique durant les dernières décennies, dans les domaines de l'informatique et de la robotique notamment, ont été si importants qu'ils permettent aujourd'hui la mise en application de systèmes complexes au sein des véhicules. Ces systèmes visent dans un premier temps à réduire les risques inhérents à la conduite en assistant les conducteurs, puis dans un second temps à offrir des moyens de transport entièrement autonomes. Les méthodes de SLAM multi-objets actuellement intégrées au sein de ces véhicules reposent pour majeure partie sur l'utilisation de capteurs embarqués très performants tels que des télémètres laser, au coût relativement élevé. Les caméras numériques en revanche, de par leur coût largement inférieur, commencent à se démocratiser sur certains véhicules de grande série et assurent généralement des fonctions d'assistance à la conduite, pour l'aide au parking ou le freinage d'urgence, par exemple. En outre, cette implantation plus courante permet également d'envisager leur utilisation afin de reconstruire l'environnement dynamique proche des véhicules en trois dimensions. D'un point de vue scientifique, les techniques de SLAM visuel multi-objets existantes peuvent être regroupées en deux catégories de méthodes. La première catégorie et plus ancienne historiquement concerne les méthodes stéréo, faisant usage de plusieurs caméras à champs recouvrants afin de reconstruire la scène dynamique observée. La plupart reposent en général sur l'utilisation de paires stéréo identiques et placées à faible distance l'une de l'autre, ce qui permet un appariement dense des points d'intérêt dans les images et l'estimation de cartes de disparités utilisées lors de la segmentation du mouvement des points reconstruits. L'autre catégorie de méthodes, dites monoculaires, ne font usage que d'une unique caméra lors du processus de reconstruction. Cela implique la compensation du mouvement propre du système d'acquisition lors de l'estimation du mouvement des autres objets mobiles de la scène de manière indépendante. Plus difficiles, ces méthodes posent plusieurs problèmes, notamment le partitionnement de l'espace de départ en plusieurs sous-espaces représentant les mouvements individuels de chaque objet mobile, mais aussi le problème d'estimation de l'échelle relative de reconstruction de ces objets lors de leur agrégation au sein de la scène statique. La problématique industrielle de cette thèse, consistant en la réutilisation des systèmes multi-caméras déjà implantés au sein des véhicules, majoritairement composés d'un caméra frontale et de caméras surround équipées d'objectifs très grand angle, a donné lieu au développement d'une méthode de reconstruction multi-objets adaptée aux systèmes multi-caméras hétérogènes en stéréo wide-baseline. Cette méthode est incrémentale et permet la reconstruction de points mobiles éparses, grâce notamment à plusieurs contraintes géométriques de segmentation des points reconstruits ainsi que de leur trajectoire. Enfin, une évaluation quantitative et qualitative des performances de la méthode a été menée sur deux jeux de données distincts, dont un a été développé durant ces travaux afin de présenter des caractéristiques similaires aux systèmes hétérogènes existants
This Ph.D. thesis, which has been carried out in the automotive industry in association with Renault Group, mainly focuses on the development of advanced driver-assistance systems and autonomous vehicles. The progress made by the scientific community during the last decades in the fields of computer science and robotics has been so important that it now enables the implementation of complex embedded systems in vehicles. These systems, primarily designed to provide assistance in simple driving scenarios and emergencies, now aim to offer fully autonomous transport. Multibody SLAM methods currently used in autonomous vehicles often rely on high-performance and expensive onboard sensors such as LIDAR systems. On the other hand, digital video cameras are much cheaper, which has led to their increased use in newer vehicles to provide driving assistance functions, such as parking assistance or emergency braking. Furthermore, this relatively common implementation now allows to consider their use in order to reconstruct the dynamic environment surrounding a vehicle in three dimensions. From a scientific point of view, existing multibody visual SLAM techniques can be divided into two categories of methods. The first and oldest category concerns stereo methods, which use several cameras with overlapping fields of view in order to reconstruct the observed dynamic scene. Most of these methods use identical stereo pairs in short baseline, which allows for the dense matching of feature points to estimate disparity maps that are then used to compute the motions of the scene. The other category concerns monocular methods, which only use one camera during the reconstruction process, meaning that they have to compensate for the ego-motion of the acquisition system in order to estimate the motion of other objects. These methods are more difficult in that they have to address several additional problems, such as motion segmentation, which consists in clustering the initial data into separate subspaces representing the individual movement of each object, but also the problem of the relative scale estimation of these objects before their aggregation within the static scene. The industrial motive for this work lies in the use of existing multi-camera systems already present in actual vehicles to perform dynamic scene reconstruction. These systems, being mostly composed of a front camera accompanied by several surround fisheye cameras in wide-baseline stereo, has led to the development of a multibody reconstruction method dedicated to such heterogeneous systems. The proposed method is incremental and allows for the reconstruction of sparse mobile points as well as their trajectory using several geometric constraints. Finally, a quantitative and qualitative evaluation conducted on two separate datasets, one of which was developed during this thesis in order to present characteristics similar to existing multi-camera systems, is provided

Local features have become an essential tool in visual recognition. Much of the progress in computer vision over the past decade has built on simple, local representations such as SIFT or HOG. SIFT in particular shifted the paradigm in feature representation. Subsequent works have often focused on improving either computational efficiency, or invariance properties. This thesis belongs to the latter group. Invariance is a particularly relevant aspect if we intend to work with dense features. The traditional approach to sparse matching is to rely on stable interest points, such as corners, where scale and orientation can be reliably estimated, enforcing invariance; dense features need to be computed on arbitrary points. Dense features have been shown to outperform sparse matching techniques in many recognition problems, and form the bulk of our work. In this thesis we present strategies to enhance low-level, local features with mid-level, global cues. We devise techniques to construct better features, and use them to handle complex ambiguities, occlusions and background changes. To deal with ambiguities, we explore the use of motion to enforce temporal consistency with optical flow priors. We also introduce a novel technique to exploit segmentation cues, and use it to extract features invariant to background variability. For this, we downplay image measurements most likely to belong to a region different from that where the descriptor is computed. In both cases we follow the same strategy: we incorporate mid-level, "big picture" information into the construction of local features, and proceed to use them in the same manner as we would the baseline features. We apply these techniques to different feature representations, including SIFT and HOG, and use them to address canonical vision problems such as stereo and object detection, demonstrating that the introduction of global cues yields consistent improvements. We prioritize solutions that are simple, general, and efficient. Our main contributions are as follows: (a) An approach to dense stereo reconstruction with spatiotemporal features, which unlike existing works remains applicable to wide baselines. (b) A technique to exploit segmentation cues to construct dense descriptors invariant to background variability, such as occlusions or background motion. (c) A technique to integrate bottom-up segmentation with recognition efficiently, amenable to sliding window detectors.
Les "features" locals s'han convertit en una eina fonamental en el camp del reconeixement visual. Gran part del progrés experimentat en el camp de la visió per computador al llarg de l'última decada es basa en representacions locals de baixa complexitat, com SIFT o HOG. SIFT, en concret, ha canviat el paradigma en representació de característiques visuals. Els treballs que l'han succeït s'acostumen a centrar o bé a millorar la seva eficiencia computacional, o bé propietats d'invariança. El treball presentat en aquesta tesi pertany al segon grup. L'invariança es un aspecte especialment rellevant quan volem treballab amb "features" denses, és a dir per a cada pixel. La manera tradicional d'atacar el problema amb "features" de baixa densitat consisteix en seleccionar punts d'interés estables, com per exemple cantonades, on l'escala i l'orientació poden ser estimades de manera robusta. Les "features" denses, per definició, han de ser calculades en punts arbitraris de la imatge. S'ha demostrat que les "features" denses obtenen millors resultats en tècniques de correspondència per a molts problemes en reconeixement, i formen la major part del nostre treball. En aquesta tesi presentem estratègies per a enriquir "features" locals de baix nivell amb "cues" o dades globals, de mitja complexitat. Dissenyem tècniques per a construïr millors "features", que usem per a atacar problemes tals com correspondències amb un grau elevat d'ambigüetat, oclusions, i canvis del fons de la imatge. Per a atacar ambigüetats, explorem l'ús del moviment per a imposar consistència espai-temporal mitjançant informació d'"optical flow". També presentem una tècnica per explotar dades de segmentació que fem servir per a extreure "features" invariants a canvis en el fons de la imatge. Aquest mètode consisteix en atenuar els components de la imatge (i per tant les "features") que probablement corresponguin a regions diferents a la del descriptor que estem calculant. En ambdós casos seguim la mateixa estratègia: la nostra voluntat és incorporar dades globals d'un nivell de complexitat mitja a la construcció de "features" locals, que procedim a utilitzar de la mateixa manera que les "features" originals. Aquestes tècniques són aplicades a diferents tipus de representacions, incloent SIFT i HOG, i mostrem com utilitzar-les per a atacar problemes fonamentals en visió per computador tals com l'estèreo i la detecció d'objectes. En aquest treball demostrem que introduïnt informació global en la construcció de "features" locals podem obtenir millores consistentment. Donem prioritat a solucions senzilles, generals i eficients. Aquestes són les principals contribucions de la tesi: (a) Una tècnica per a reconstrucció estèreo densa mitjançant "features" espai-temporals, amb l'avantatge respecte a treballs existents que podem aplicar-la a càmeres en qualsevol configuració geomètrica ("wide-baseline"). (b) Una tècnica per a explotar dades de segmentació dins la construcció de descriptors densos, fent-los invariants a canvis al fons de la imatge, i per tant a problemes com les oclusions en estèreo o objectes en moviment. (c) Una tècnica per a integrar segmentació de manera ascendent ("bottom-up") en problemes de reconeixement d'una manera eficient, dissenyada per a detectors de tipus "sliding window".

The SIMBIO-SYS instrument suite of the BepiColombo ESA mission to Mercury is equipped with the Stereo Imaging Channel (STC) stereo telescope, based on an innovative and compact design, in which two independent optical channels oriented at ±20° with respect to Nadir collect light on a common detector. The stereo acquisition mode is based on the push frame concept, which has never been adopted on space missions before. To demonstrate and characterize the capability of the instrument to reconstruct a tridimensional surface with the desired accuracy by means of the stereo push-frame concept, an innovative experimental setup has been realized. The problem of working at an essentially infinite object distance over hundreds km baselines has been overcome by means of a simple collimator lens and two precision rotation stages, scaling down the stereo reconstruction problem in terms of baseline and accuracy requirements. The stereo validation has been performed by comparing the shape of a target object accurately measured by laser scanning, with the shape reconstructed by applying classical stereo algorithms to the acquired image pairs. The reconstruction of depth information from stereo images of an observed surface requires the characterization of the imaging system in terms of camera calibration process. The particular application required a preliminary analysis of the most suited camera model by verifying the calibration procedure performance. To this end, an innovative method has been developed for the simulation of the calibration data and for the evaluation of calibration algorithms. A preliminary test of the stereo validation procedure has been performed with an evaluation model of STC. The obtained results show the goodness of this innovative technique, that will be applied also for validating the stereo capabilities of STC flight model.
La suite di strumenti SIMBIO-SYS, a bordo della missione ESA BepiColombo per l’esplorazione del pianeta Mercurio, sarà equipaggiata con un canale per ricostruzione stereo denominato Stereo Imaging Channel (STC). Il telescopio è caratterizzato da un design compatto e innovativo in cui due sotto-canali, orientati a ±20°rispetto a Nadir, convogliano la luce catturata su di un unico sensore. L’acquisizione delle immagini avverrà in modalità push-frame, mai impiegata finora nell’ambito di missioni spaziali. La necessità di verificare e caratterizzare prima del lancio le capacità dello strumento di effettuare una ricostruzione tridimensionale della superficie del pianeta con l’accuratezza richiesta ha portato alla progettazione e allo sviluppo di un apparato e di un metodo sperimentale innovativo. La baseline dello strumento è stata riprodotta in scala in laboratorio introducendo alcuni semplici componenti, tra cui: una lente, che funge da collimatore, e due rotatori ad alta precisione. L’impiego di un collimatore, oltre a comportare una riduzione della distanza di osservazione da alcune centinaia di chilometri a circa un metro, implica un adeguamento dei requisiti in termini di risoluzione e accuratezza della ricostruzione. L’apparato opto-meccanico realizzato ha consentito l’acquisizione in laboratorio di immagini stereoscopiche di una superficie di prova: l’attività di validazione stereo si è svolta mettendo a confronto una mappa tridimensionale di riferimento dell’oggetto, prodotta da un laser scanner sufficientemente accurato, col modello ottenuto applicando tecniche di ricostruzione stereo a coppie di immagini del campione stesso. La ricostruzione di informazione circa la profondità della superficie osservata a partire da coppie stereo di immagini richiede che il sistema venga caratterizzato tramite una procedura di calibrazione. Vista la particolare applicazione, è stato sviluppato un metodo grazie al quale è stato possibile simulare i dati di calibrazione, con lo scopo di valutare l’efficacia e l’accuratezza di tecniche di calibrazione classiche applicate al particolare caso di studio. E’ stato effettuato un test preliminare applicando il setup e la procedura di validazione stereo a un modello funzionale di STC: i risultati ottenuti hanno dimostrato l’efficacia dell’apparato per la validazione stereo, che verrà applicato al modello di volo dello strumento.

Cette thèse étudie la précision en vision stéréo, les méthodes de détection dites a contrario et en présente une application à l'imagerie satellitaire. La première partie a été réalisée dans le cadre du projet DGA-ANR-ASTRID "STÉRÉO". Son but est de définir les limites effectives des méthodes de reconstruction stéréo quand on contrôle toute la chaîne d’acquisition à la précision maximale, que l’on acquiert des paires stéréo en rapport B/H très faible et sans bruit. Pour valider ce concept, nous créons des vérités terrains très précises en utilisant un rendeur. En gardant les rayons calculés durant le rendu, nous avons une information très dense sur la scène 3D. Ainsi nous créons des cartes d'occultations, de disparités dont l'erreur de précision est inférieure à 10e-6. Nous avons mis à la disposition de la communauté de recherche des images de synthèse avec un SNR supérieur à 500 : un ensemble de 66 paires stéréo dont le B/H varie de1/2500 à 1/50. Pour évaluer les méthodes de stéréo sur ce nouveau type de données, nous proposons des métriques calculant la qualité des cartes de disparités estimées, combinant la précision et la densité des points dont l'erreur relative est inférieure à un certain seuil. Nous évaluons plusieurs algorithmes représentatifs de l'état de l'art, sur les paires créées ainsi sur les paires de Middlebury, jusqu'à leurs limites de fonctionnement. Nous confirmons par ces analyses, que les hypothèses théoriques sur le bien-fondé du faible B/H en fort SNR sont valides, jusqu'à une certaine limite que nous caractérisons. Nous découvrons ainsi que de simples méthodes de flux optique pour l'appariement stéréo deviennent plus performantes que des méthodes variationnelles discrètes plus élaborées. Cette conclusion n'est toutefois valide que pour des forts rapports signal à bruit. L'exploitation des données denses nous permet de compléter les vérités terrain par une détection très précise des bords d'occultation. Nous proposons une méthode de calcul de contours vectoriels subpixéliens à partir d'un nuage de points très dense, basée sur des méthodes a contrario de classification de pixels. La seconde partie de la thèse est dédiée à une application du flot optique subpixélien et des méthodes a contrario pour détecter des nuages en imagerie satellitaire. Nous proposons une méthode qui n'exploite que l'information visible optique. Elle repose sur la redondance temporelle obtenue grâce au passage répété des satellites au-dessus des mêmes zones géographiques. Nous définissons quatre indices pour séparer les nuages du paysage : le mouvement apparent inter-canaux, la texture locale, l'émergence temporelle et la luminance. Ces indices sont modélisés dans le cadre statistique des méthodes a contrario qui produisent un NFA (nombre de fausses alarmes pour chacun). Nous proposons une méthode pour combiner ces indices et calculer un NFA beaucoup plus discriminant. Nous comparons les cartes de nuages estimées à des vérités terrain annotées et aux cartes nuageuses produites par les algorithmes liés aux satellites Landsat-8 etSentinel-2. Nous montrons que les scores de détection et de fausses alarmes sont supérieurs à ceux obtenus avec ces algorithmes, qui pourtant utilisent une dizaine de bandes multi-spectrales
This thesis studies the accuracy in stereo vision, detection methods calleda contrario and presents an application to satellite imagery. The first part was carried out within the framework of the project DGA-ANR-ASTRID"STEREO". His The aim is to define the effective limits of stereo reconstruction when controlling the entire acquisition chain at the maximum precision, that one acquires stereo pairs in very low baseline and noise-free. To validate thisconcept, we create very precise ground truths using a renderer. By keeping the rays computed during rendering, we have very dense information on the 3Dscene. Thus we create occultation and disparity maps whose precision error is less than 10e-6. We have made synthetic images available to the research community with an SNR greater than 500: a set of 66 stereo pairs whoseB/H varies from 1/2500 to 1/50. To evaluate stereo methods onthis new type of data, we propose metrics computing the quality of the estimated disparity maps, combining the precision and the density of the points whose relative error is less than a certain threshold. We evaluate several algorithmsrepresentative of the state of the art, on the pairs thus created and on theMiddlebury pairs, up to their operating limits. We confirm by these analyzesthat the theoretical assumptions about the merit of the low B/H in highSNR are valid, up to a certain limit that we characterize. We thus discover that simple optical flow methods for stereo matching become more efficient than more sophisticated discrete variational methods. This conclusion, however, is only valid for high signal-to-noise ratios. The use of the dense data allows us to complete the ground truths a subpixel detection of the occlusion edges. We propose a method to compute subpixel vector contours from a very dense cloud ofpoints, based on pixel classification a contrario methods. The second part of the thesis is devoted to an application of the subpixelian optical flowand a contrario methods to detect clouds in satellite imagery. We propose a method that exploits only visible optical information. It is based onthe temporal redundancy obtained by the repeated passages of the satellites overthe same geographical zones. We define four clues to separate the clouds fromthe landscape: the apparent inter-channel movement, Local texture, temporal emergence and luminance. These indices are modeled in the statistical framework of a contrario methods which produce an NFA (number of false alarms for each). We propose a method for combining these indices and computing a much more discriminating NFA. We compare the estimated cloud maps to annotated ground truths and the cloud maps produced by the algorithms related to the Landsat-8and Sentinel-2 satellites. We show that the detection and false alarms scores are higher than those obtained with these algorithms, which however use a dozen multi-spectral bands

Perception of depth information is central to three-dimensional (3D) vision problems. Stereopsis is an important passive vision technique for depth perception. Wide baseline stereo is a challenging problem that attracts much interest recently from both the theoretical and application perspectives. In this research we approach the problem of wide baseline stereo using the geometric and structural constraints within feature sets. The major contribution of this dissertation is that we proposed and implemented a more efficient paradigm to handle the challenges introduced by perspective distortion in wide baseline stereo, compared to the state-of-the-art. To facilitate the paradigm, a new feature-matching algorithm that extends the state-of-the-art matching methods to larger baseline cases is proposed. The proposed matching algorithm takes advantage of both the local feature descriptor and the structure pattern of the feature set, and enhances the matching results in the case of large viewpoint change. In addition, an innovative rectification for uncalibrated images is proposed to make wide baseline stereo dense matching possible. We noticed that present rectification methods did not take into account the need for shape adjustment. By introducing the geometric constraints of the pattern of the feature points, we propose a rectification method that maximizes the structure congruency based on Delaunay triangulation nets and thus avoid some existing problems of other methods. The rectified stereo images can then be used to generate a dense depth map of the scene. The task is much simplified compared to some existing method because the 2D searching problem is reduced to 1D searching. To validate the proposed methods, real world images are applied to test the performance and comparisons to the state-of-the-art methods are provided. The performance of the dense matching with respect to the changing baseline is also studied.

碩士
國立交通大學
電子研究所
101
In this thesis, we present a wide-baseline stereo system for 3D scene reconstruction. We implement our system with multiple un-calibrated cameras which are set widely. The main challenge of the system lies on how to match image pairs at wide-baseline, in which there appear large perspective distortions and large occlusion areas between images. In this research, we attempt to tackle the problem based on machine learning and optimization techniques. In order to match image more accurately, we apply random forest to overcome large perspective distortions, and add Conditional Random Field (CRF) with modified Histogram of Oriented Gradients (HOG) to solve the matching problem. Combining conditional random field with random forest can not only correct error correspondences but handle some occlusions. After getting matched points, we use these correspondences to find a 3D point set and camera matric by bundle adjustment (BA) that minimizes re-projection error. Then, we use the idea of spectral matting to refine the 3D point set. Finally, we build a 3D model with the refined point set.

This research work is organized in two parts. The first one addresses how to analyze the interdisciplinary problem of motorcycle safety, in order to maximize the positive impact of future motorcycle safety systems through a new methodological tool, which it was developed in this work. The second part describes a remote sensor, developed in this work, aimed at avoiding or mitigating the motorcycle crashes, as these were found necessary in the first part. Therefore, this research focused on the requirements to accomplish a conceptual safety functionality, called Motorcycle Autonomous Emergency Braking (M-AEB). The reason for this is because this safety system does not exist for motorcycles. The specifications behind the M-AEB functionality were taken from the outcomes of a prior EU research project called ABRAM (Autonomous BRAking for Motorcycles).

The aim of this thesis is to analyse and interpret the existent and future needs of the people and environment of the mission station, Kwasizabantu,to ensure settlement growth that will enhance the social, economical and environmental aspects of the settlement. This will result in a development framework of the whole site, a master plan for the lifespan of the settlement and detailed design of the heart of the settlement.
Dissertation (ML(Prof))--University of Pretoria, 2009.
Architecture
unrestricted