Much research effort has been devoted to producing algorithms that contribute directly or indirectly to the extraction of 3D information from a wide variety of types of scenes and conditions of image capture. The research work presented in this thesis is aimed at three distinct applications in this area: interactively extracting 3D points from a pair of uncalibrated images in a flexible way; finding corresponding points automatically in high resolution images, particularly those of archaeological scenes captured from a freely moving light aircraft; and improving a correlation approach to dense disparity mapping leading to 3D surface reconstructions. The fundamental concepts required to describe the principles of stereo vision, the camera models, and the epipolar geometry described by the fundamental matrix are introduced, followed by a detailed literature review of existing methods. An interactive system for viewing a scene via a monochrome or colour anaglyph is presented which allows the user to choose the level of compromise between amount of colour and ghosting perceived by controlling colour saturation, and to choose the depth plane of interest. An improved method of extracting 3D coordinates from disparity values when there is significant error is presented. Interactive methods, while very flexible, require significant effort from the user finding and fusing corresponding points and the thesis continues by presenting several variants of existing scale invariant feature transform methods to automatically find correspondences in uncalibrated high resolution aerial images with improved speed and memory requirements. In addition, a contribution to estimating lens distortion correction by a Levenberg Marquard based method is presented; generating data strings for straight lines which are essential input for estimating lens distortion correction. The remainder of the thesis presents correlation based methods for generating dense disparity maps based on single and multiple image rectifications using sets of automatically found correspondences and demonstrates improvements obtained using the latter method. Some example views of point clouds for 3D surfaces produced from pairs of uncalibrated images using the methods presented in the thesis are included. / Al-Baath University / The appendices files and images are not available online.

No description available.

In this dissertation we introduce two variable selection procedures for multivariate responses. Our procedures are based on sufficient dimension reduction concepts and are model-free. In the first procedure we consider the dual marginal coordinate hypotheses, where the role of the predictor and the response is not important. Motivated by canonical correlation analysis (CCA), we propose a CCA-based test for the dual marginal coordinate hypotheses, and devise a joint backward selection algorithm for dual model-free variable selection. The second procedure is based on ordinary least squares (OLS). We derive and study the asymptotic properties of the OLS-based test under the normality assumption of the predictors as well as an asymmetry assumption. When these assumptions are violated, the asymptotic test with elliptical trimming and clustering is still valid with desirable numerical performances. A backward selection algorithm for the predictor is also provided for the OLS-based test. The performances of the proposed tests and the variable selection procedures are evaluated through synthetic examples and a real data analysis. / Statistics

Geodetiska provfält är viktiga för att geodetiska instrument kan kontrolleras och kalibreras. Exempelvis så kan en längdbas användas för att kontrollera nollpunktsfelet hos en Electronic Distance Measuement. Provfältets höga noggrannhet möjliggör även att detaljpunkterna exempelvis kan användas för att kontrollera noggrannheten vid utsättning eller inmätning. Att hålla provfältets koordinater uppdaterade är viktigt för att bibehålla den höga noggrannheten. Tidigare forskning visar att kända punkter kan vara bra för att kontrollera GNSS-antenner innan de sätts i bruk. Precisionen av mätningar var något man valde att utveckla på Statens Institution för Byggnadsforskning i Gävle på 70-talet (Statens Institut för Byggnadsforskning, 1984). Byggforskningen byggde ett provfält i Mårtsbo där punkter mättes in som sedan redovisats i en koordinatlista. Till en början användes provfältet för att kontrollera olika instrument och mätmetoder, men nu har det stått orört under en längre tid. Denna studie görs för att ta reda på om punkterna i området har rört sig något sen 1977 då fältet upprättades och därigenom avgöra om punkterna måste tilldelas nya koordinater innan användningen av fältet återupptas.   Innan mätning kontrollerades och kalibrerades instrumenten som skulle användas. För att koppla stomnätet till SWEREF 99 16 30 och RH 2000 gjordes en statisk GNSS-mätning på de två mest öppna punkterna i nätet. Vinklar och avstånd mellan punkterna erhölls genom att använda satsmätning med multistation. För att få med alla detaljpunkter gjordes en koordinattransformation med hjälp av de gamla koordinaterna och de nya som mätts in. I efterarbetet bearbetades GNSS-mätningarna i programvaran Leica Geo Office. Satsmätningarna behandlades i programvaran SBG Geo och där gjordes två nätutjämningar, en med lokala koordinater och en med koordinater i  SWEREF 99 16 30 och RH 2000. Slutligen transformerades resterande koordinater från det lokala systemet till SWEREF 99 16 30 och RH 2000 i  programvaran Gtrans. Baserat på resultatet har signifikant deformation uppstått på flera av punkterna. De största förändringarna som skett är i plan 2,2 mm i P5 och i höjd 1,7 mm i P7. Noggrannheten som uppnåtts i denna studie är inte lika bra som tidigare, men ändå tillräckligt för att använda fältet. / Geodetic test fields are important because they allow geodetic instruments to be tested and calibrated. For example, the baseline can be used when checking the zero-point error in an Electronic Distance Measurement. The high accuracy of the test field will also make it possible to use the detail points for checking the accuracy of the measurements when surveying. Keeping the coordinates of the field updated is important to maintain the high accuracy of the points. Previous research shows that known points were used to check GNSS-antennas before commercial use. The precision of measurements was something that the Swedish Institute of Constructional Science chose to develop even further in the late 1970s. A geodetic test field was constructed in Mårtsbo where points were established and presented in a coordinate list. The field was at first used to control different measuring equipment and methods, but in later years it has not been used at all. The aim of this study is to re-measure and update the coordinates of the test field. For this study, the instruments was controlled and calibrated before any measurements. To link to the well-established points with known coordinates in the control network static GNSS-measurements were performed on the two points located in the most open areas and positioned in SWEREF 99 16 30 and RH 2000. The angles and distances between the points were given by using rounds of measurements. To get coordinates for every detail point a coordinate transformation was used with the measured points as control points. During the post-processing of the GNSS-data the Leica Geo Office software was used to compute the coordinates. The rounds of measurements with multistation were processed in the software SBG Geo and network adjustments with both old and new coordinates were carried out. Finally, the rest of the coordinates were transformed into the new coordinate system which was done in the software Gtrans.. The result shows that significant local deformation has occurred on several points. The largest of deformation in plane is 2,2 mm in P5 and in height is 1,7 mm in P7. The uncertainties of the measured points are a bit higher in this study compared to earlier one, but they are still good enough to be use in this test field.

Due to modularity and easy scalability, modular multilevel converters (MMCs) are deemed the most suitable for high-voltage and medium-voltage power conversion applications. However, large module capacitors are usually required in MMCs to store large circulating power of line-frequency and its harmonics that flow through the capacitors. Even though several methods for minimizing the circulating power have been proposed in the literature, there is still the need for a systematic and simplified approach of addressing these control strategies and evaluating their efficacy. Moreover, the generally accepted feedback control architecture for the MMC is complicated, derived through a rigorous mathematical analysis, and therefore, not easy to intuitively comprehend. Recently, a method of modeling of the MMC based on state-plane analysis and coordinate transformation, is proposed in the literature. Based on the state-plane analysis, two kinds of circulating power in the MMC are identified that are orthogonal to each other. This means these two circulating power can be controlled individually without affecting each other. To control these circulating power, in the literature, a decoupled equivalent circuit model is developed through the coordinate transformation which clearly suggests a means for minimizing these circulating power. Further extending this work, in this thesis, the existing control concepts for reducing the circulating power are unveiled in a systematic and simplified manner utilizing the decoupled equivalent circuit model. A graphical visualization of circulating power using the state-planes is provided for each control strategy to readily compare its efficacy. Moreover, the generally accepted control architecture of the MMC is presented in an intuitive and simplified way using the decoupled circuit model. The important physics related to control implementation, originally hidden behind the complicated mathematics, is explained in detail. / Master of Science / A power converter is an electrical device that converts electrical energy from one form to another in order to be compatible with the load demand. A typical power converter consists of semiconductor switches, inductor, capacitor etc. These power converters are required in a wide range of applications: automotive and traction, motor drives, renewable energy conversion, energy storage, aircraft, power generation, transmission, and distribution, to name a few. Many of these applications are continuously increasing their power capacity to handle the escalating demands of energy that exist due to rising population numbers, industrialization, urbanization etc. Consequently, it has been a responsibility of power electronics engineers and researchers to develop power converters that can handle high voltages and high currents. Multilevel power converters have been the key-enabling developments that can withstand high-voltages while using traditional low-voltage semiconductor switches. Several multilevel converters such as the neutral point clamped converter, flying capacitor converter, cascaded H-bridge converter, modular multilevel converter (MMC) etc. have been developed and commercialized in the last two decades. Among them, the MMC is a widely accepted topology for medium- and high-voltage power conversion applications. In an MMC, several modules are stacked together in series, and each module consists of semiconductor switches and a capacitor. The series connection of the modules enables the MMC to handle high-voltage power conversion using low-voltage traditional semiconductor switches. The voltage rating of an MMC can be easily scaled-up by simply increasing the number of modules in each arm. Moreover, since several identical modules are connected in each arm, the structure of the MMC is highly modular which helps greatly in manufacturing and design. Nonetheless, in MMCs, generally large circulating power flow to the capacitor in each module, which leads to significant voltage ripples. To suppress these voltage ripples, a large capacitor is required in each module, leading to large size and weight of the converter. In the literature, several control strategies have been proposed to minimize the circulating power. However, there is still the need for a systematic and simplified approach of addressing these control strategies and evaluating their efficacy. Moreover, the generally accepted feedback control architecture for the MMC is complicated, derived through a rigorous mathematical analysis, and therefore, not easy to intuitively comprehend. Recently, a decoupled equivalent circuit model has been developed in the literature. This model clearly explains the process of power flow in the MMC between input and output and the nature of the circulating power. The equivalent circuit model provides the circulating power, that are orthogonal to each other, meaning they can be controlled individually without affecting each other. Moreover, the equivalent circuit model clearly suggests a means for minimize the circulating power by providing two "ideal" control laws. Further extending this work, in this thesis, the existing control concepts for reducing the circulating power are unveiled in a systematic and simplified manner utilizing the decoupled equivalent circuit model. Moreover, the generally accepted control architecture of the MMC is presented in an intuitive and simplified way via the decoupled circuit model. The important physics related to control implementation, originally hidden behind the complicated mathematics, is explained in detail.

京都大学 / 新制・課程博士 / 博士(工学) / 甲第25237号 / 工博第5196号 / 新制||工||1992(附属図書館) / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 原田 英治, 教授 後藤 仁志, 准教授 音田 慎一郎, 教授 山上 路生 / 学位規則第4条第1項該当 / Doctor of Agricultural Science / Kyoto University / DFAM

Im Rahmen einer Diplomarbeit wurden 2015 die untertägigen Polygonzugnetze der Gruben Unterbreizbach/Merkers der K+S KALI GmbH im Lagebezug ETRS89/UTM neu orientiert und Ergebnisse verschiedener Messepochen ausgeglichen. Für die Schaffung eines weiteren Anschlusspunktes erfolgten zur Koordinatenübertragung im Schacht Merkers 2 eine mechanische Schachtlotung und eine Steilschachtmessung nach dem Verfahren von Mecke/Ackermann aus dem Jahr 2002. Die Methoden werden gegenübergestellt und die erreichten Genauigkeiten diskutiert. / Within the framework of a diploma thesis the underground mining coordinate system of Unterbreizbach/Merkers K+S KALI GmbH became reorientated in the reference system ETRS89/UTM with an adjustment of results from different ages of measurement. For the development of a further termination point a mechanical shaft plumbing and a steep shaft measurement according to Mecke/Ackermann from 2002 took place. The methods will be compared followed by a discussion of the reached accuracies.

Splitter plates are passive flow control devices for reducing drag and suppressing vortex shedding from bluff bodies. Most studies of splitter plates involve the flow around an “infinite” circular cylinder, however, in the present study the flow around a surface-mounted finite-height circular cylinder, with a wake-mounted splitter plate, was studied experimentally in a low-speed wind tunnel using a force balance and single-component hot-wire anemometry. Four circular cylinders of aspect ratios AR = 9, 7, 5 and 3 were tested for a Reynolds number range of Re = 1.9×10^4 to 8.2×10^4. The splitter plates had lengths, relative to the cylinder diameter, of L/D = 1, 1.5, 2, 3, 5 and 7, thicknesses ranging from T/D = 0.10 and 0.15, and were the same height as the cylinder being tested. The cylinders were partially immersed in a flat-plate turbulent boundary layer, where the range of boundary layer thickness relative to the cylinder diameter was δ/D = 1.4 to 1.5. Measurements were made of the mean drag force coefficient, the Strouhal number at the mid-height position, and the Strouhal number and power spectra along the cylinder height. For all four finite circular cylinders, the splitter plates were effective at reducing the magnitude of the Strouhal number, and weakening or even suppressing vortex shedding, depending on the specific combination of AR and L/D. Compared to the case of an infinite circular cylinder, the splitter plate is less effective at reducing the mean drag force coefficient of a finite circular cylinder. The largest drag reduction was obtained for the cylinder of AR = 9 and splitter plates of L/D = 1 to 3, while negligible drag reduction occurred for the shorter cylinders.

Le modèle computo-fonctionnel de la vision de haut niveau de Kosslyn et Koenig (1992) repose sur la dissociation entre la reconnaissance et le traitement spatial de la scène visuelle. En 1987, Kosslyn postula l’existence de deux processus pour le traitement des relations spatiales : un processus catégoriel qui calcule les positions relatives des objets et un processus coordonné qui calcule la distance entre les objets. Des études utilisant le paradigme de présentation en champ visuel divisé ont mis en évidence que l’hémisphère gauche sous-tendrait un traitement catégoriel ; l’hémisphère droit sous-tendrait un traitement coordonné. Cette interaction semblerait valider la dichotomie des deux types de processus.Une revue de la littérature pluridisciplinaire a été menée afin de savoir si ce fait est suffisant pour rejeter l’hypothèse d’un processus unique pour les traitements catégoriels et coordonnés. Entre autres, plusieurs études ont observé un effet de pratique au cours de la réalisation de jugements coordonnés : une diminution de l’intervention de l’hémisphère droit au profit d’une prise en charge progressive de l’hémisphère gauche. De plus, l’avantage de l’hémisphère gauche pour le traitement catégoriel a été rarement observé.Une série de cinq expériences comportementales ont été conduites pour vérifier certains biais expérimentaux qui pourraient expliquer les différences hémisphériques et l’effet de pratique observés. Nos résultats nous ont permis d’avancer des hypothèses axées sur la communication entre les hémisphères et sur un lien entre les processus catégoriels et coordonnés. Une critique du paradigme de présentation en champ visuel divisé et différents modèles d’interaction hémisphérique ont été présentés. Trois expériences ont été menées, afin d’évaluer l’impact des communications hémisphériques dans le traitement des relations spatiales. A la lumière de nos résultats, la dichotomie des processus catégoriels et coordonnés a été discutée. / The computational-functional conception of high-level processing of vision in Kosslyn and Koenig (1992) relies on dissociation between object recognition and spatial processing. In 1987, Kosslyn postulated that two different processes compute spatial-relations: categorical process computes relative position of objects and coordinate process computes the distance between objects. Some studies indicate a left-hemisphere advantage for processing categorical spatial relations and a right-hemisphere advantage for processing coordinate spatial relations. This hemispheric difference is interpreted as an evidence of a dichotomy between these two processes. A pluridisciplinary review was conducted to assure that single process hypothesis is dismissed out. Some studies showed, in a coordinate task, that practice resulted in a decreased right-hemisphere involvement and a concurrent increase in left-hemisphere involvement (i.e., practice effect). Furthermore, the left-hemisphere advantage in categorical was seldom observed. The theoretical aim of the thesis was based on two questions: (1) Are there single or several processes encoding visual spatial relations? (2) How can we interpret this practice effect ?Five experiments were run to verify possible bias which might explain observed hemispheric differences and practice effect. Our results suggested that hemispheric communication might be an important factor in spatial-relation processing.An theoretical investigation of divided visual field paradigm was led and several interhemispheric models were described. Three experiments were conducted to estimate hemispheric communication in spatial-relation process. In the light of our results, separate categorical-coordinate processes hypothesis were discussed.

La géométrie d'une pièce fabriquée réelle diffère de la pièce virtuelle de CAO (Conception Assistée par Ordinateur. Cette différence est due à la somme des écarts inhérents à la fabrication. L'objectif de ce travail est d’introduire des pièces virtuelles ayant des défauts de forme (Skin Model Shape) dans les applications d'ingénierie afin de répondre aux exigences croissantes de l'industrie en matière de gestion de la qualité de la géométrie des produits. Les travaux traitent de divers aspects, particulièrement de la génération de défauts de forme, de la simulation d'assemblage et de la métrologie virtuelle.Les méthodes permettant de générer des défauts de forme sur des surfaces simples sont analysés et classées. En raison des défauts de forme, la combinaison de surfaces simples pour générer une pièce entière induit une incohérence géométrique au niveau des arêtes. Une méthode globale basée sur les éléments finis et une méthode locale basée sur le lissage local de maillage sont utilisées pour résoudre ce problème.Pour prédire l'écart des caractéristiques fonctionnelles, la simulation d'assemblage est effectuée en utilisant des surfaces avec défauts de forme. Une approche est développée sur la base de la condition de complémentarité linéaire et du torseur de petits déplacements pour prendre en compte les conditions aux limites de l'assemblage, telles que les déplacements et les charges.Des méthodes pour évaluer les écarts sur les modèles de surfaces avec défauts de forme sont également étudiées. Les spécifications sur le produit sont exprimées avec GeoSpelling et évaluées à l'aide du torseur de petits déplacements. Les méthodes développées sont intégrées dans un laboratoire virtuel pour l'apprentissage en ligne.Les études susmentionnées complètent et étendent les méthodes de gestion des tolérances basées sur GeoSpelling et le « skin » modèle. / The geometry of a real manufactured part differs from the virtual workpieces designed in Computer Aided Design (CAD) systems. This difference is due to the accumulation of unavoidable manufacturing deviations. The objective of this work is to implement virtual workpieces with form defects (Skin Model Shape) in engineering applications to meet the industry’s increasing demands in product geometry quality management. Various aspects are covered here, in particular form defect generation, assembly simulation and virtual metrology.Methods to generate form defects on simple surfaces are reviewed and classified. Due to form defects, the combination of simple surfaces to generate a whole part led to inconsistency on the edges. A global FEA-based method and a local mesh smoothing based method are used to overcome this issue.To predict the deviation of functional characteristics, assembly simulation is conducted using skin model shapes. An approach is developed based on the Linear Complementarity Condition and the Small Displacement Torsor to take into account assembly boundary conditions, such as displacements and loads.Methods to evaluate deviation values on skin model shapes are also studied. Product specifications are expressed with GeoSpelling, and evaluated using the Small Displacement Torsor method. The developed methods are integrated into an online Virtual Laboratory for e-learning.The above-mentioned studies complement and extend the tolerance management methods based on GeoSpelling and skin models.