[pt] Imprescindível para quaisquer aplicações de visão computacional ou realidade aumentada, a calibração de câmera é o processo no qual se obtém os parâmetros intrínsecos e extrínsecos da câmera, tais como distância focal, ponto principal e valores que mensuram a distorção ótica da lente. Atualmente o método mais utilizado para calibrar uma câmera envolve o uso de imagens de um padrão planar em diferentes perspectivas, a partir das quais se extrai pontos de controle para montar um sistema de equações lineares cuja solução representa os parâmetros da câmera, que são otimizados com base no erro de reprojeção 2D. Neste trabalho, foi escolhido o padrão de calibração aneliforme por oferecer maior precisão na detecção dos pontos de controle. Ao aplicarmos técnicas como transformação frontal-paralela, refinamento iterativo dos pontos de controle e segmentação adaptativa de elipses, nossa abordagem apresentou melhoria no resultado do processo de calibração. Além disso, propomos estender o modelo de otimização ao redefinir a função objetivo, considerando não somente o erro de reprojeção 2D, mas também o erro de colinearidade 2D. / [en] Crucial for any computer vision or augmented reality application, the camera calibration is the process in which one gets the intrinsics and the extrinsics parameters of a camera, such as focal length, principal point and distortions values. Nowadays, the most used method to deploy the calibration comprises the use of images of a planar pattern in different perspectives, in order to extract control points to set up a system of linear equations whose solution represents the camera parameters, followed by an optimization based on the 2D reprojection error. In this work, the ring calibration pattern was chosen because it offers higher accuracy on the detection of control points. Upon application of techniques such as fronto-parallel transformation, iterative refinement of the control points and adaptative segmentation of ellipses, our approach has reached improvements in the result of the calibration process. Furthermore, we proposed extend the optimization model by modifying the objective function, regarding not only the 2D reprojection error but also the 2D collinearity error.

[pt] Câmeras de pixel único consistem em reconstruir computacionalmente imagens em duas dimensões a partir de um conjunto de medidas feitas por um detector de único pixel. Para que se obtenha a informação espacial, um conjunto de padrões de modulação são aplicados à luz transmitida/refletida do objeto e essa informação é combinada com o sinal integral do detector. Primeiro, apresentamos uma visão geral desses sistemas e demonstramos a implementação de uma prova de conceito capaz de fazer aquisição de imagem usando três modos de operação: Varredura, escaneamento por base de Hadamard, e escaneamento por base de Hadamard com sensoriamento compreensivo. Segundo, discutimos como os diferentes parâmetros experimentais do sistema ótico afetam a aquisição. Finalmente, comparamos a performance dos três modos de operação quando usados para a aquisição de images com tamanhos entre (8px, 8px) e (128px, 128px). / [en] Single-pixel imaging consists in computationally reconstructing 2-dimensional images from a set of intensity measurements taken by a singlepoint detector. To derive the spatial information of a scene, a set of modulation patterns are applied to the transmitted/backscattered light from the object and combined with the integral signal on the detector. First, we present an overview of such optical systems and implement a proof of concept that can perform image acquisition using three different modes of operation: Raster scanning, Hadamard basis scanning, and Hadamard compressive sampling. Second, we explore how the different experimental parameters affect image acquisition. Finally, we compare how the three scanning mode perform for acquisition of images of sizes ranging from (8px, 8px) to (128px, 128px).

Walking and cycling are considered very important behaviours in public health. This thesis is a study of the measurement of walking, cycling and overall travel behaviour. In the first part I present a systematic review of studies comparing Global Positioning System (GPS) measured travel to self-report. I found 12 results from eight eligible studies. All studies showed self-reported journey times were greater than GPS measured times. The differences ranged from over-reporting of +2.2 to +13.5 minutes per journey. In the second part of this thesis, I describe the development of a feasible, valid and reliable method for measuring travel with automated wearable cameras (SenseCam) through field testing and two pilot studies. I compared my new method to direct observation (considered a criterion measure) and found very good agreement and reliability (inter-rater, intra-rater and inter-measure). I also present an ethical framework for the measurement of health-related behaviours using automated wearable cameras. In the third part of the thesis, I report findings from an experiment designed to assess a well known UK travel diary (The National Travel Survey). Across four locations (Oxford, UK; Romford, UK; San Diego, USA; and Auckland, New Zealand) I collected 3-4 days of SenseCam and travel diary data from n=84 participants (convenience sample). Compliance with the collection protocol was high and inspection of the crude results suggests relative agreement between measures. Analysis of matched pairs of measurements (n=1,127 journeys) suggests a significant positive bias on self-reported durations (2:08 minutes; 95% CI = 1:48 to 2:28; 95% limits-of-agreement = -9:10 to 13:26). These results suggest self-reported journey exposure is valid at a population level, though corrections according to my reported bias could be considered. The large limits of agreement on duration estimates suggest self-report may be unsuitable for assessment of individual travel behaviour.

PEPT (positron emission particle tracking) is a technique for tracking a small radioactive tracer in Lagrangian co-ordinates. The technique was used to study the flow patterns of non-Newtonian CMC (Carboxy Methyl Cellulose) solutions inside a vessel agitated by an axial flow impeller. The 'non-intrusive' PEPT technique uses two position-sensitive detectors to track a radioactive particle in space and time. The particle is labelled with a positron emitting isotope. Once emitted from the nucleus a positron annihilates with an electron releasing energy in the form of two 511 keV back-to-back gamma-rays travelling in opposite directions, 180 degrees apart. The tracer particle is introduced into the stirred vessel which is mounted between the two detectors of the positron camera. Three axial flow impellers produced by Lightnin Mixers Ltd were used to carry out the experiments. Results showed that the discharge from the three impellers was radial when agitating non-Newtonian viscous solutions of CMC. Trajectory analysis was used to compare the performance of the impellers using the agitation index and the efficiency of circulation. A limited number of experiments was carried out to compare the effect of baffles on the circulation of the fluids in a mixing tank. The results showed that mixing of these non-Newtonian liquids in an unbaffled tank is better than in a baffled tank when using axial flow impellers. Other experiments were carried out to suspend solid particles in viscous fluids. Results showed that the minimum speed required to suspend large particles is lower than that required to suspend small particles. There are many correlations and models in the literature to determine the minimum speed required to suspend all the particles in a fluid; some of these correlations and models were compared with experimental results from this work. The correlation of Zweitering (1958) agreed with experimental data after modification. The Geisler et al. (1993) model agreed with the data provided that the power consumption is correctly substituted. The last part of this work concerned the flow of non-Newtonian viscous materials through industrial equipment. Yoghurt was chosen as the test fluid as one of the companies sponsoring this project was Eden Vale, a yoghurt manufacturer. A method was proposed using rheological measurements to simulate the flow through the dispensing pipeline and distributing nozzles; this method allows the designer to predict the final properties of yoghurt after passing through the paching head. Measurements were also carried out to determine the final gel structure of yoghurt in the delivery pots. This data of this thesis is useful in designing stirred tanks when non-Newtonian fluid is present, either for agitation or when suspending solids. Also, a method was provided to design yoghurt manufacturing line.

Thesis (MSc)--Stellenbosch University, 2014. / ENGLISH ABSTRACT: Monocular structure from motion (SfM) is a widely researched problem, however many of the existing approaches prove to be too computationally expensive for use on mobile devices. In this thesis we investigate how inertial sensors can be used to increase the performance of SfM algorithms on mobile devices. Making use of the low cost inertial sensors found on most mobile devices we design and implement an extended Kalman filter (EKF) to exploit their complementary nature, in order to produce an accurate estimate of the attitude of the device. We make use of a quaternion based system model in order to linearise the measurement stage of the EKF, thus reducing its computational complexity. We use this attitude estimate to enhance the feature tracking and camera localisation stages in our SfM pipeline. In order to perform feature tracking we implement a hybrid tracking algorithm which makes use of Harris corners and an approximate nearest neighbour search to reduce the search space for possible correspondences. We increase the robustness of this approach by using inertial information to compensate for inter-frame camera rotation. We further develop an efficient bundle adjustment algorithm which only optimises the pose of the previous three key frames and the 3D map points common between at least two of these frames. We implement an optimisation based localisation algorithm which makes use of our EKF attitude estimate and the tracked features, in order to estimate the pose of the device relative to the 3D map points. This optimisation is performed in two steps, the first of which optimises only the translation and the second optimises the full pose. We integrate the aforementioned three sub-systems into an inertial assisted pose estimation pipeline. We evaluate our algorithms with the use of datasets captured on the iPhone 5 in the presence of a Vicon motion capture system for ground truth data. We find that our EKF can estimate the device’s attitude with an average dynamic accuracy of ±5°. Furthermore, we find that the inclusion of sensors into the visual pose estimation pipeline can lead to improvements in terms of robustness and computational efficiency of the algorithms and are unlikely to negatively affect the accuracy of such a system. Even though we managed to reduce execution time dramatically, compared to typical existing techniques, our full system is found to still be too computationally expensive for real-time performance and currently runs at 3 frames per second, however the ever improving computational power of mobile devices and our described future work will lead to improved performance. From this study we conclude that inertial sensors make a valuable addition into a visual pose estimation pipeline implemented on a mobile device. / AFRIKAANSE OPSOMMING: Enkel-kamera struktuur-vanaf-beweging (structure from motion, SfM) is ’n bekende navorsingsprobleem, maar baie van die bestaande benaderings is te berekeningsintensief vir gebruik op mobiele toestelle. In hierdie tesis ondersoek ons hoe traagheidsensors gebruik kan word om die prestasie van SfM algoritmes op mobiele toestelle te verbeter. Om van die lae-koste traagheidsensors wat op meeste mobiele toestelle gevind word gebruik te maak, ontwerp en implementeer ons ’n uitgebreide Kalman filter (extended Kalman filter, EKF) om hul komplementêre geaardhede te ontgin, en sodoende ’n akkurate skatting van die toestel se postuur te verkry. Ons maak van ’n kwaternioon-gebaseerde stelselmodel gebruik om die meetstadium van die EKF te lineariseer, en so die berekeningskompleksiteit te verminder. Hierdie afskatting van die toestel se postuur word gebruik om die fases van kenmerkvolging en kameralokalisering in ons SfM proses te verbeter. Vir kenmerkvolging implementeer ons ’n hibriede volgingsalgoritme wat gebruik maak van Harris-hoekpunte en ’n benaderde naaste-buurpunt-soektog om die soekruimte vir moontlike ooreenstemmings te verklein. Ons verhoog die robuustheid van hierdie benadering, deur traagheidsinligting te gebruik om vir kamerarotasies tussen raampies te kompenseer. Verder ontwikkel ons ’n doeltreffende bondelaanpassingsalgoritme wat slegs optimeer oor die vorige drie sleutelraampies, en die 3D punte gemeenskaplik tussen minstens twee van hierdie raampies. Ons implementeer ’n optimeringsgebaseerde lokaliseringsalgoritme, wat gebruik maak van ons EKF se postuurafskatting en die gevolgde kenmerke, om die posisie en oriëntasie van die toestel relatief tot die 3D punte in die kaart af te skat. Die optimering word in twee stappe uitgevoer: eerstens net oor die kamera se translasie, en tweedens oor beide die translasie en rotasie. Ons integreer die bogenoemde drie sub-stelsels in ’n pyplyn vir postuurafskatting met behulp van traagheidsensors. Ons evalueer ons algoritmes met die gebruik van datastelle wat met ’n iPhone 5 opgeneem is, terwyl dit in die teenwoordigheid van ’n Vicon bewegingsvasleggingstelsel was (vir die gelyktydige opneming van korrekte postuurdata). Ons vind dat die EKF die toestel se postuur kan afskat met ’n gemiddelde dinamiese akkuraatheid van ±5°. Verder vind ons dat die insluiting van sensors in die visuele postuurafskattingspyplyn kan lei tot verbeterings in terme van die robuustheid en berekeningsdoeltreffendheid van die algoritmes, en dat dit waarskynlik nie die akkuraatheid van so ’n stelsel negatief beïnvloed nie. Al het ons die uitvoertyd drasties verminder (in vergelyking met tipiese bestaande tegnieke) is ons volledige stelsel steeds te berekeningsintensief vir intydse verwerking op ’n mobiele toestel en hardloop tans teen 3 raampies per sekonde. Die voortdurende verbetering van mobiele toestelle se berekeningskrag en die toekomstige werk wat ons beskryf sal egter lei tot ’n verbetering in prestasie. Uit hierdie studie kan ons aflei dat traagheidsensors ’n waardevolle toevoeging tot ’n visuele postuurafskattingspyplyn kan maak.

published_or_final_version / abstract / Electrical and Electronic Engineering / Doctoral / Doctor of Philosophy

published_or_final_version / abstract / Electrical and Electronic Engineering / Master / Master of Philosophy

Project PANOPTES (http://www.projectranoptes.org) is aimed at establishing a collaboration between professional astronomers, citizen scientists and schools to discover a large number of exoplanets with the transit technique. We have developed digital camera based imaging units to cover large parts of the sky and look for exoplanet transits. Each unit costs approximately $5000 USD and runs automatically every night. By using low-cost, commercial digital single-lens reflex (DSLR) cameras, we have developed a uniquely cost-efficient system for wide field astronomical imaging, offering approximately two orders of magnitude better etendue per unit of cost than professional wide-field surveys. Both science and outreach, our vision is to have thousands of these units built by schools and citizen scientists gathering data, making this project the most productive exoplanet discovery machine in the world.

La tomoscintigraphie myocardique est considérée comme un examen de référence pour le diagnostic et l'évaluation de la maladie coronarienne. Mise au point dans les années 1980, cette technique est en pleine mutation depuis l'arrivée de nouvelles caméras à semi-conducteurs. Deux caméras à semi-conducteurs, dédiées à la cardiologie nucléaire et utilisant des détecteurs de Cadmium Zinc Telluride sont actuellement commercialisées : la Discovery NM-530c (General Electric) et la DSPECT (Spectrum Dynamics). Les performances de ces caméras CZT ont été évaluées : 1) à la fois sur fantôme et sur des examens d'effort provenant de patients à faible probabilité de maladie coronaire, et 2) avec les paramètres d'acquisition et de reconstruction utilisés en clinique. Les résultats ont démontré la nette supériorité des caméras CZT en termes de sensibilité de détection, de résolution spatiale et de rapport contraste sur bruit par rapport à la génération de caméras d'Anger. Ces propriétés vont permettre de diminuer très fortement les temps d'acquisition et les activités injectées, tout en améliorant la qualité des images. Néanmoins, on connaît encore mal les limites et possibles artéfacts liés à la géométrie particulière d'acquisition. C'est pourquoi nous avons développé, avec la plateforme de simulations Monte Carlo GATE, un simulateur numérique spécifique de la caméra DSPECT. Nous avons pu ensuite le valider en comparant des données effectivement enregistrées aux données simulées. Ce simulateur pourrait aider à optimiser les protocoles de reconstruction et d'acquisition, en particulier les protocoles les plus complexes (acquisitions double traceur, études cinétiques) / Myocardial single-photon emission computed tomography (SPECT) is considered as the gold standard for the diagnosis of coronary artery disease. Developed in the 1980s with rotating Anger gamma-cameras, this technique could be dramatically enhanced by new imaging systems working with semi-conductor detectors. Two semiconductor cameras, dedicated to nuclear cardiology and equipped with Cadmium Zinc Telluride detectors, have been recently commercialized: the Discovery NM- 530c (General Electric) and the DSPECT (Spectrum Dynamics). The performances of these CZT cameras were compared: 1) by a comprehensive analysis of phantom and human SPECT images considered as normal and 2) with the parameters commonly recommended for SPECT recording and reconstruction. The results show the superiority of the CZT cameras in terms of detection sensitivity, spatial resolution and contrast-to-noise ratio, compared to conventional Anger cameras. These properties might lead to dramatically reduce acquisition times and/or the injected activities. However, the limits of these new CZT cameras, as well as the mechanism of certain artefacts, remain poorly known. That?s why we developed, with the GATE Monte Carlo simulation plateform, a specific simulator of the DSPECT camera. We validated this simulator by comparing actually recorded data with simulated data. This simulator may yet be used to optimize the recorded and reconstruction processes, especially for complex protocols such as simultaneous dual-radionuclide acquisition and kinetics first-pass studies

"Hearing loss affects approximately 1 in 10 people in the world and this percentage is increasing every year. Some of the most common causes of hearing loss are disorders of the middle-ear. Early detection and diagnosis of hearing loss as well as research to understand the hearing processes depend on medical and research tools for quantification of hearing capabilities and the function of the middle-ear in the complex acousto-mechanical transformation of environmental sounds into vibrations of the middle-ear, particular of the human tympanic membrane (TM or eardrum). Current ear exams assess the state of a patient’s hearing capabilities mainly based on qualitative evaluation of the healthiness of the TM. Existing quantitative clinical methods for description of the motion of the TM are limited to either average acoustic estimates (admittance or reflectance) or single-point displacement measurements. Such methods could leave examiners and researchers blind to the complex spatio-temporal response of the nanometer scale displacements of the entire TM. Current state-of-the-art medical research tools provide full-field nanometer displacement measurements of the surface of the human TM excited by steady state (tonal) stimuli. However, to fully understand the mechanics of hearing, and the complex acousto-mechanical characteristics of TM in particular, new tools are needed for full-field high-speed characterization of the nanometer scale displacements of the human TM subjected to impulse (wideband) acoustic excitation. This Dissertation reports the development of a new high-speed holographic system (HHS) for full-field nanometer transient (i.e., > 10 kHz) displacement measurement of the human middle-ear and the tympanic membrane, in particular. The HHS allows spatial (i.e., >500k data points) and temporal (i.e., > 40 kHz) resolutions that enable the study of the acoustical and mechanical characteristics of the middle-ear at a level of detail that have never been reached before. The realization of the HHS includes the development and implementation of novel phase sampling and acquisition approaches that allow the use of state-of-the-art high-resolution (i.e., >5 MP) and high-speed (> 80,000 fps) cameras through modular and expandable control architectures. The development of novel acquisition approaches allows the use of conventional speed (i.e., <20 fps) cameras to realize high-temporal resolutions (i.e., <15 us) at equivalent sampling rates of > 50,000 fps with minimum hardware cost and modifications. The design and implementation of novel spatio-temporal phase sampling methods utilize the high temporal resolution (i.e., < 5 us exposure) and frame rate (i.e., >80,000 fps) of high-speed cameras without imposing constraints on their spatial resolution (i.e., >20 um pixel size). Additionally, the research and in-vivo applications capabilities of the HHS are extended through the development and implementation of a holographic otoscope head (OH) and a mechatronic otoscope positioner (MOP). The large (i.e., > 1 GB with > 8x10^9 parameters) spatio-temporal data sets of the HHS measurements are automatically processed by custom parallel data mining and interpretation (PDMI) methods, which allow automatic quantification of medically relevant motion parameters (MRMPs), such as modal frequencies, time constants, and acoustic delays. Such capabilities could allow inferring local material properties across the surface of the TM. The HHS is a new medical tool that enables otologists to improve the quality of diagnosis and treatments as well as provides researchers with spatio-temporal information of the hearing process at a level of detail never reached before. "