This thesis describes reconstruction of scene which is scan trough two cameras. There are described methods of calibration of cameras system, methods for finding the corners and methods for finding correspondences. Corners are searched by FAST detector and for search correspondences are used normalized cross correlation. In the framework of 3D reconstruction is implemented rectification. The final shape is saved to VRML format. In the thesis are described parallelization options. The calculation of the correlation is optimized for multiprocessors CPU and there are designed implementations of algorithm to GPU and FPGA too.

This thesis deals with 3D reconstruction using stereo cameras. This work is to show the usefulness of GPU acceleration for sophisticated algorithm

Laboratory mouse is a appropriate experimental model for studies on normal tooth development mechanisms and for understanding of etio-pathogenesis of dental anomalies in humans. Mouse dentition consists of one incisor separated from three molars by toothless diastema in each jaw quadrant. Although an adult mice diastema does not contain teeth, rudimentary tooth primordia (MS, R2) appear here during embryonic development. However, their development stops and consequently functional tooth does not arise here. It is known that in mice with mutations in Spry2 and Spry4 genes, supernumerary teeth arise in antemolar region. The understanding of the temporo-spatial dynamics of signaling centers in Spry2/Spry4 double transgenic mice may help to elucidate the role of these embryonic diastemal buds in supernumerary tooth formation. The research of possible regeneration of dental anomalies in off-spring of crossing Spry4 deficient mice with supernumerary teeth formation with Eda deficient mice with predisposition to reduced dentition could be the way to elucidating of mechanism of dental pathologies development in general. Our aim was to describe temporal and spatial dynamics of Shh signaling domains in Spry2/Spry4 embryos, because Shh is one of the markers of early odontogenesis and determining of...

No description available.

<p>A previously designed lens system is constructed and characterized for possible application to engine diagnostics and 3D reconstructions of combusting flows. Since optical engines cannot be operated at the same conditions as a production engine, optical access to a production engine, with minimum changes in the engine geometry is necessary for developing a better understanding of the in-cylinder processes. The application of a probe designed to fit in the pressure transducer port of a diesel engine was demonstrated in this work. Measurement of various optical parameters established the ability of the lens system to have a good resolution over the entire stroke length of the cylinder. The temperature analysis of the probe and lens system confirmed its ability to withstand the high-temperature conditions in the engine cylinder head. The lens system design was coupled with imaging fibers to transmit images from the image plane of the probe to a high-speed camera located at a safe distance from the combusting environment. Due to the robust design of the probe and its compatibility with an imaging fiber, the probe was identified to be a good alternative as an inexpensive lens system for tomography in challenging environments. To validate its use, 3D reconstruction of a sewing pin using a range of views was demonstrated. Parameters affecting the reconstructions were identified and optimized to obtain high-quality reconstructions. </p>

<p>3D character modeling is a crucial process of asset creation in the entertainment industry, particularly for animation and games. A fully automated pipeline via sketch-based 3D modeling (SBM) is an emerging possibility, but development is stalled by unrefined outputs and a lack of character-centered tools. This thesis proposes an improved method for constructing 3D character models with minimal user input, using only two sketch inputs  i.e., a front and side unshaded sketch. The system implements a deep convolutional neural network (CNN), a type of deep learning algorithm extending from artificial intelligence (AI), to process the input sketch and generate multi-view depth, normal and confidence maps that offer more information about the 3D surface. These are then fused into a 3D point cloud, which is a type of object representation for 3D space. This point cloud is converted into a 3D mesh via an occupancy network, involving another CNN, for a more precise 3D representation. This reconstruction step contends with non-deep learning approaches such as  Poisson reconstruction. The proposed system is evaluated for character generation on standardized quantitative metrics (i.e., Chamfer Distance [CD], Earth Mover’s Distance [EMD], F-score and Intersection of Union [IoU]), and compared to the base framework trained on the same character sketch and model database. This implementation offers a  significant improvement in the accuracy of vertex positions for the reconstructed character models. </p>

No description available.

Virtuelle 3D-Modelle finden in den historischen Disziplinen in zweierlei Art Verwendung. Zum einen dienen diese zur Erfassung und Digitalisierung existierender historischer Objekte. Daneben dient die Erstellung von virtuellen 3D-Rekonstruktionen der Nachbildung nicht mehr existierender Objekte und Strukturen. Während technische Abläufe beider Ansätze ebenso wie methodische und wissenschaftstheoretische Aspekte in der Fachliteratur häufig und umfassend thematisiert werden, ist eine Frage nach sozialen Aspekten und sozialer Interaktion im Kontext derartiger Vorhaben bisher unbeleuchtet geblieben. Ziel der Arbeit stellt dar, eine Bandbreite und Relevanz von Aspekten der Kooperation als „Zusammenarbeit mit gemeinsamem Ziel, gegenseitiger Abstimmung, planvollem Vorgehen sowie Vorteilen für alle Akteure“ (Hagenhoff, 2004) im Kontext derartiger geschichtswissenschaftlicher 3D-Modellierungsvorhaben mittels sozialwissenschaftlicher Methoden zu beleuchten. Dabei zielt eine Darlegung auf unterschiedliche Skalierungen von Kooperation ab – angefangen bei einer Wissenschaftslandschaft über Kooperationsstrukturen bis hin zu einer Betrachtung von spezifischen Kooperationsphänomenen und -strategien innerhalb von Arbeits- und Erstellungsprozessen.

Virtuelle 3D-Modelle finden in den historischen Disziplinen in zweierlei Art Verwendung. Zum einen dienen diese zur Erfassung und Digitalisierung existierender historischer Objekte. Daneben dient die Erstellung von virtuellen 3D-Rekonstruktionen der Nachbildung nicht mehr existierender Objekte und Strukturen. Während technische Abläufe beider Ansätze ebenso wie methodische und wissenschaftstheoretische Aspekte in der Fachliteratur häufig und umfassend thematisiert werden, ist eine Frage nach sozialen Aspekten und sozialer Interaktion im Kontext derartiger Vorhaben bisher unbeleuchtet geblieben. Ziel der Arbeit stellt dar, eine Bandbreite und Relevanz von Aspekten der Kooperation als „Zusammenarbeit mit gemeinsamem Ziel, gegenseitiger Abstimmung, planvollem Vorgehen sowie Vorteilen für alle Akteure“ (Hagenhoff, 2004) im Kontext derartiger geschichtswissenschaftlicher 3D-Modellierungsvorhaben mittels sozialwissenschaftlicher Methoden zu beleuchten. Dabei zielt eine Darlegung auf unterschiedliche Skalierungen von Kooperation ab – angefangen bei einer Wissenschaftslandschaft über Kooperationsstrukturen bis hin zu einer Betrachtung von spezifischen Kooperationsphänomenen und -strategien innerhalb von Arbeits- und Erstellungsprozessen.

Purpose – We propose a multidisciplinary approach based on an extensive data base which provides digitalized photographic material from the end of the 19th century up to recent times. Thus a large amount of photographic evidence will be exploited, structured and enriched by additional sources to serve as a foundation for an application relying on 3D visualizations. The application addresses scholars as well as the general public and will provide different kinds of information and tools for research and knowledge transfer. Design/methodology/approach – The method applied will be diachronic: the virtual model may show one point in urban history depicting a certain state of past Dresden and also its development through the various eras. In addition the method works in a dualistic mode: on the one hand the physical development of the urban area will be explored and presented in detail, on the other hand the analysis of the pictures will give profound insights in the specific perception of the urban space. Originality/value – This methodology aims to make large repositories more accessible and proactive in information-seeking. Using a 3D application as an access for media repositories, research tools and functionalities which can improve the scientific handling of the data will be considered. How should the data and information be processed to meet the researcher’s needs? Which information can be retrieved from the visual media? What needs to be considered to ensure scientific standards and motivation while working with the image repositories? Users of the virtual archives can benefit extensively form effective searching functions and tools which work not only content- and theme-based but also location-based. Practical implications – The outcomes of the research will be presented in a 4D browser and available in an Augmented Reality presentation. The design will comply with the requirements of the field of application, whether aiming at a scientific, educative or touristic purpose. The paper itself considers three different approaches to the topic highlighting the multidisciplinary strategy and opportunities of the project. The first one considers research questions from art history. The second one reflects on concepts from information science, photogrammetry and computer vision for visualizations and the third one introduces an interaction concept for an AR application for the Zwinger in Dresden.