In this thesis, an innovative algorithm for improving the accuracy of variational space-time stereoscopic reconstruction of ocean surfaces is presented. The space-time reconstruction method, developed based on stereo computer vision principles and variational optimization theory, takes videos captured by synchronized cameras as inputs and produces the shape and superficial pattern of an overlapped region of interest as outputs. These outputs are designed to be the minimizers of the variational optimization framework and are dependent on the estimation of the camera parameters. Therefore, from the perspective of computer vision, the proposed algorithm adjusts the estimation of camera parameters to lower the disagreement between the reconstruction and 2-D camera recordings. From a mathematical perspective, since the minimizers of the variational framework are determined by a set of partial differential equations (PDEs), the algorithm modifies the coefficients of the PDEs based on the current numerical solutions to reduce the minimum of the optimization framework. Our algorithm increases the tolerance to the errors of camera parameters, so the joint operations of our algorithm and the variational reconstruction method can generate accurate space-time models even using videos captured by perturbed cameras as input. This breakthrough prompts the realization of ocean surface reconstruction using videos filmed by remotely-controlled helicopters in the future. A number of techniques, technical or theoretical, are explored to fulfill the development and implementation of the algorithm and relative computation issues. The effectiveness of the proposed algorithm is validated through the statistics applied to real ocean surface reconstructions of data collected from an offshore platform at the Crimean Peninsula, the Black Sea. Moreover, synthetic data generated using computer graphics are customized to simulate various situations that are not recorded in the Crimea dataset for the demonstration of the algorithm.

In general, stereovision can be defined as a two part problem. The first is the correspondence problem. This involves determining the image point in each image of a set of images that correspond to the same physical point P. We will call this set of image points, N. The second problem is the reconstruction problem. Once a set of image points, N, that correspond to point P has been determined, N is then used to extract three dimensional information about point P. This master's thesis presents three novel solutions to the reconstruction problem. Two of the techniques presented are for detecting the location of a 3-D point and one for detecting a line expressed in a three dimensional coordinate system. These techniques are tested and validated using a unique 3-D finger detection algorithm. The techniques presented are unique because of their simplicity and because they do not require the cameras to be placed in specific locations, orientations or have specific alignments. On the contrary, it will be shown that the techniques presented in this thesis allow the two cameras used to assume almost any relative pose provided that the object of interest is within their field of view. The relative pose of the cameras at a given instant in time, along with basic equations from the perspective image model are used to form a system of equations that when solved, reveal the 3-D coordinates of a particular fixed point of interest or the three dimensional equation of a fixed line of interest. Finally, it will be shown that a single moving camera can successfully perform the same line and point detection accomplished by two cameras by altering the pose of the camera. The results presented in this work are beneficial to any typical stereovision application because of the computational ease in comparison to other point and line reconstruction techniques. But more importantly, this work allows for a single moving camera to perceive three-dimensional position information, which effectively removes the two camera constraint for a stereo vision system. When used with other monocular cues such as texture or color, the work presented in this thesis could be as accurate as binocular stereo vision at interpreting three dimensional information. Thus, this work could potentially increase the three dimensional perception of a robot that normally uses one camera, such as an eye-in-hand robot or a snake like robot.

Among the various approaches to 3-D imaging, structured light illumination (SLI) is widely spread. SLI employs a pair of digital projector and digital camera such that the correspondences can be found based upon the projecting and capturing of a group of designed light patterns. As an active sensing method, SLI is known for its robustness and high accuracy. In this dissertation, I study the phase shifting method (PSM), which is one of the most employed strategy in SLI. And, three novel approaches in PSM have been proposed in this dissertation. First, by regarding the design of patterns as placing points in an N-dimensional space, I take the phase measuring profilometry (PMP) as an example and propose the edge-pattern strategy which achieves maximum signal to noise ratio (SNR) for the projected patterns. Second, I develop a novel period information embedded pattern strategy for fast, reliable 3-D data acquisition and reconstruction. The proposed period coded phase shifting strategy removes the depth ambiguity associated with traditional phase shifting patterns without reducing phase accuracy or increasing the number of projected patterns. Thus, it can be employed for high accuracy realtime 3-D system. Then, I propose a hybrid approach for high quality 3-D reconstructions with only a small number of illumination patterns by maximizing the use of correspondence information from the phase, texture, and modulation data derived from multi-view, PMP-based, SLI images, without rigorously synchronizing the cameras and projectors and calibrating the device gammas. Experimental results demonstrate the advantages of the proposed novel strategies for 3-D SLI systems.

No / An algorithm for the rectification of uncalibrated images is presented and applied to a variety of cases. The algorithm generates the rectifying transformations directly from the geometrical relationship between the images, using any three correspondences in the images to define a reference plane. A small set of correspondences is used to calculate an initial rectification. Additional correspondences are introduced semi-automatically, by correlating regions of the rectified images. Since the rectified images of surfaces in the reference plane have no relative distortion, features can be matched very accurately by correlation, allowing small changes in disparity to be detected. In the 3-d reconstruction of an architectural scene, differences in depth are resolved to about 0.001 of the distance from camera to subject.

No description available.

Structured light illumination (SLI) is the process of projecting a series of light striped patterns such that, when viewed at an angle, a digital camera can reconstruct a 3-D model of a target object's surface. But by relying on a series of time multiplexed patterns, SLI is not typically associated with video applications. For this purpose of acquiring 3-D video, a common SLI technique is to drive the projector/camera pair at very high frame rates such that any object's motion is small over the pattern set. But at these high frame rates, the speed at which the incoming video can be processed becomes an issue. So much so that many video-based SLI systems record camera frames to memory and then apply off-line processing. In order to overcome this processing bottleneck and produce 3-D point clouds in real-time, we present a lookup-table (LUT) based solution that in our experiments, using a 640 by 480 video stream, can generate intermediate phase data at 1063.8 frames per second and full 3-D coordinate point clouds at 228.3 frames per second. These achievements are 25 and 10 times faster than previously reported studies. At the same time, a novel dual-frequency pattern is developed which combines a high-frequency sinusoid component with a unit-frequency sinusoid component, where the high-frequency component is used to generate robust phase information and the unit-frequency component is used to reduce phase unwrapping ambiguities. Finally, we developed a gamma model for SLI, which can correct the non-linear distortion caused by the optical devices. For three-step phase measuring profilometry (PMP), analysis of the root mean squared error of the corrected phase showed a 60х reduction in phase error when the gamma calibration is performed versus 33х reduction without calibration.

Dans le circuit de la greffe de cornée, l'endothélium de chaque greffon est observé en microscopie optique conventionnelle afin de vérifier que sa densité cellulaire est suffisante pour maintenir une bonne transparence après l'opération. Les greffons étant conservés dans un milieu spécifique, ils sont imprégnés de liquide et présentent donc des plis qui perturbent l'observation et le comptage des cellules. Ce problème pratique est à l'origine d’une étude théorique sur les concepts de profondeur de champ étendue et de shape-from-focus. A partir d'une séquence d'images acquise par sectionnement optique, les informations les plus nettes permettent d'une part d'accéder à la topographie de la surface observée et d'autre part de restaurer l'image de sa texture. Une reconstruction surfacique 3-D est alors obtenue en projetant la texture sur la topographie. Cette thèse considère essentiellement l’étape fondamentale de mesure de netteté du processus de reconstruction. Des nouvelles mesures génériques offrant une haute sensibilité à la netteté sont introduites. De par une stratégie 3-D originale au travers de la séquence d'images, une autre mesure très robuste au bruit est proposée. Toutes ces mesures sont testées sur des données simulées puis diverses acquisitions réelles en microscopie optique conventionnelle et comparées aux méthodes de la littérature. Par ailleurs, la mesure 3-D améliore nettement les reconstructions d'endothéliums cornéens à partir de leurs acquisitions particulièrement perturbées (inversions de contraste). Un processus itératif complet de reconstruction 3-D d’endothéliums cornéens est finalement décrit, aboutissant à des résultats solides et exploitables. / In the cornea transplant process, each graft endothelium is observed by conventional optical microscopy to check that its cell density is sufficient to maintain a proper transparency after the transplantation. The grafts are stored in a specific preservation medium, they are thus impregnated with fluid and therefore exhibit folds which make cell observation and counting difficult. This practical issue led to the following theoretical study about the so-called concepts: extended-depth-of-field and shape-from-focus. Throughout a sequence of images acquired by optical sectioning, the in-focus information allows on the one hand to recover the topography of the observed surface and on the other hand to restore the image of its texture. A 3-D reconstruction is then obtained by mapping the texture onto the topography. This thesis basically considers the fundamental step of the reconstruction process that is the focus measurement. New generic focus measurements exhibiting high sharpness sensitivity are introduced. Another one offering high noise robustness is proposed, due to an original 3-D strategy through the image sequence, unlike traditional methods that operate in 2-D. All of them are tested on simulated data and various real acquisitions, and compared to the state-of-the-art methods. Furthermore, the aforementioned 3-D focus measurement clearly improves the 3-D surface reconstructions of the corneal endotheliums from their particularly disturbed acquisitions (contrast reversals). A complete iterative process of 3-D reconstruction of the corneal endothelial surfaces is finally described, resulting in solid results that can already be transferred to cornea banks.

L’acétylcholine (ACh) endogène de la corne dorsale de la moelle épinière (CDME) exerce une analgésie puissante utilisée en clinique, dont la source et les mécanismes demeurent inconnus. Elle siège probablement au niveau d’un plexus de fibres cholinergiques de la CDME d’origine non-élucidée. Dans ce contexte, nous avons pu établir que ce plexus est principalement issu d’interneurones cholinergiques spinaux caractérisés dans ces travaux, qui seraient donc le substrat probable de l’analgésie décrite. Décrits comme concourant aux effets aigus et analgésiques de la morphine, nous avons, par ailleurs, pu observer que les récepteurs de l’ACh participaient également aux effets chroniques et pro-algésique de la morphine, notamment au niveau de la CDME. Ceci place donc l’ACh comme un effecteur ou intermédiaire de la morphine.Nos travaux suggèrent ainsi que le système cholinergique spinal pourrait constituer une cible thérapeutique alternative pour de nouveaux traitements de la douleur / In the spinal cord dorsal horn (SCDH), endogenous acetylcholine (ACh) acts as a powerful analgesia, of clinical use. Though its source and mechanisms remain unravelled, this analgesia probably lies in a plexus of cholinergic fibers (PCF) located in the SCDH and of undetermined origin. In this context, we established that the PCF mainly originates from a spinal population of cholinergic interneurons, fully characterized in this work. These are, thus, the likely substrate of the spinal cholinergic analgesia.Besides, ACh receptors (AChR) partly mediate the analgesic acute effects of morphine. In this work, we also observed that a chronically-administered AChR agonist reproduces as well the pro-algesic effects of morphine in the same conditions. Thus, ACh appears as a possible intermediary or a final effecter of the morphine pain pathways.Our data suggest that the cholinergic system could become a new putative therapeutic target in pain management and treatment.