Refine
Year of publication
- 2018 (29) (remove)
Document Type
- Bachelor Thesis (19)
- Master's Thesis (4)
- Doctoral Thesis (3)
- Part of Periodical (3)
Keywords
- Bildverarbeitung (2)
- Astrophysik (1)
- Augmented Reality (1)
- Bildanalyse (1)
- Computer Vision (1)
- Computergrafik (1)
- Computervisualistik (1)
- Deep Learning (1)
- Deep Metric Learning (1)
- Extended Reality (1)
- Fell (1)
- Fell Rendering (1)
- Fluidsimulation (1)
- GPGPU (1)
- Haare (1)
- Innenarchitektur (1)
- Klassifikation (1)
- Machine Learning (1)
- Medizinische Bildverarbeitung (1)
- Mikroskopie (1)
- Multiple Object Tracking (1)
- Musik (1)
- Mustererkennung (1)
- Photogrammetrie Kalibrierung EMVA-1288 (1)
- Photogrammetry (1)
- Random Finite Sets (1)
- Rendering (1)
- Segmentierung (1)
- Sterne (1)
- Virtual Reality (1)
- fur rendering (1)
Institute
- Institut für Computervisualistik (29) (remove)
The present thesis gives an overview of the general conditions for the programming of graphics cards. For this purpose, the most important Application Programming Interfaces (APIs) available on the market are presented and compared. Subsequently, two standard algorithms from the field data processing, prefix sum and radixsort are presented and examined with regard to the implementation with parallel programming on the GPU. Both algorithms were implemented using the OpenGL-API and OpenGL compute shaders. Finally, the execution times of the two algorithms were compared.
This thesis addresses the automated identification and localization of a time-varying number of objects in a stream of sensor data. The problem is challenging due to its combinatorial nature: If the number of objects is unknown, the number of possible object trajectories grows exponentially with the number of observations. Random finite sets are a relatively new theory that has been developed to derive at principled and efficient approximations. It is based around set-valued random variables that contain an unknown number of elements which appear in arbitrary order and are themselves random. While extensively studied in theory, random finite sets have not yet become a leading paradigm in practical computer vision and robotics applications. This thesis explores random finite sets in visual tracking applications. The first method developed in this thesis combines set-valued recursive filtering with global optimization. The problem is approached in a min-cost flow network formulation, which has become a standard inference framework for multiple object tracking due to its efficiency and optimality. A main limitation of this formulation is a restriction to unary and pairwise cost terms. This circumstance makes integration of higher-order motion models challenging. The method developed in this thesis approaches this limitation by application of a Probability Hypothesis Density filter. The Probability Hypothesis Density filter was the first practically implemented state estimator based on random finite sets. It circumvents the combinatorial nature of data association itself by propagation of an object density measure that can be computed efficiently, without maintaining explicit trajectory hypotheses. In this work, the filter recursion is used to augment measurements with an additional hidden kinematic state to be used for construction of more informed flow network cost terms, e.g., based on linear motion models. The method is evaluated on public benchmarks where a considerate improvement is achieved compared to network flow formulations that are based on static features alone, such as distance between detections and appearance similarity. A second part of this thesis focuses on the related task of detecting and tracking a single robot operator in crowded environments. Different from the conventional multiple object tracking scenario, the tracked individual can leave the scene and later reappear after a longer period of absence. Therefore, a re-identification component is required that picks up the track on reentrance. Based on random finite sets, the Bernoulli filter is an optimal Bayes filter that provides a natural representation for this type of problem. In this work, it is shown how the Bernoulli filter can be combined with a Probability Hypothesis Density filter to track operator and non-operators simultaneously. The method is evaluated on a publicly available multiple object tracking dataset as well as on custom sequences that are specific to the targeted application. Experiments show reliable tracking in crowded scenes and robust re-identification after long term occlusion. Finally, a third part of this thesis focuses on appearance modeling as an essential aspect of any method that is applied to visual object tracking scenarios. Therefore, a feature representation that is robust to pose variations and changing lighting conditions is learned offline, before the actual tracking application. This thesis proposes a joint classification and metric learning objective where a deep convolutional neural network is trained to identify the individuals in the training set. At test time, the final classification layer can be stripped from the network and appearance similarity can be queried using cosine distance in representation space. This framework represents an alternative to direct metric learning objectives that have required sophisticated pair or triplet sampling strategies in the past. The method is evaluated on two large scale person re-identification datasets where competitive results are achieved overall. In particular, the proposed method better generalizes to the test set compared to a network trained with the well-established triplet loss.
In dieser Arbeit wird ein System zur Erzeugung und Darstellung stereoskopischen Video-Panoramen vorgestellt. Neben der theoretischen Grundlagen werden der Aufbau und die Funktionsweise dieses Systems erläutert.
Dazu werden spezielle Kameras verwendet, die Panoramen aufnehmen
können und zur Wiedergabe synchronisiert werden. Anschließend wird ein Renderer implementiert, welcher die Panoramen mithilfe einer VirtualReality Brille stereoskopisch darstellen kann. Dafür werden separate Aufnahmen für die beiden Augen gemacht und getrennt wiedergegeben. Zum Abschluss wird das entstandene Video-Panorama mit einem Panorama eines schon bestehenden Systems verglichen.
Mit der Microsoft Kinect waren die ersten Aufnahmen von synchronisierten Farb- und Tiefendaten (RGB-D) möglich, ohne hohe finanzielle Mittel aufwenden zu müssen und neue Möglichkeiten der Forschung eröffneten sich. Mit fortschreitender Technik sind auch mobile Endgeräte in der Lage, immer mehr zu leisten. Lenovo und Asus bieten die ersten kommerziell erwerblichen Geräte mit RGB D-Wahrnehmung an. Mit integrierten Funktionen der Lokalisierung, Umgebungserkennung und Tiefenwahrnehmung durch die Plattform Tango von Google gibt es bereits die ersten Tests in verschiedenen Bereichen des Rechnersehens z.B. Mapping. In dieser Arbeit wird betrachtet, inwiefern sich ein Tango Gerät für die Objekterkennung eignet. Aus den Ausgangsdaten des Tango Geräts werden RGB D-Daten extrahiert und für die Objekterkennung verarbeitet. Es wird ein Überblick über den aktuellen Stand der Forschung und gewisse Grundlagen bezüglich der Tango Plattform gegeben. Dabei werden existierende Ansätze und Methoden für eine Objekterkennung auf mobilen Endgeräten untersucht. Die Implementation der Erkennung wird anhand einer selbst erstellten Datenbank von RGB-D Bildern gelernt und getestet. Neben der Vorstellung der Ergebnisse werden Verbesserungen und Erweiterungen für die Erkennung vorgeschlagen.
The following work describes the prototypical conception and development of the stat-raising game "Adventurer's Guild" using the game engine Ren'Py. The game's narrative is influenced by player decisions and the planning of activities. The game is to be visually pleasing and enjoyable.
After giving an overview of stat-raising as a genre, the existing games "Dandelion - Wishes Brought to You", "Pastry Lovers", "Long Live the Queen" and "Magical Diary" are analysed to pinpoint various strengths and weaknesses of their different takes on the genre.
The resultant findings are used for the conception of a new stat-raising game.
The game mechanics and the design decisions made are then shown in screenshots and thoroughly explained.
In a final assessment, the game will be examined with regard to the given task. Further possibilities for potential improvements and expansions will be detailed at the end.
This bachelor thesis’s objective is to offer the reader insight into the discrete Fourier transform, the discrete cosine transform and the discrete Hadamard-Walsh transform in the context of image processing, and also to compare these transformations under various aspects. For this purpose the term of transformation, originated in linear algebra, will be explained and applied to image processing. Subsequently, the understanding of the Fourier transform will successively be built up and connected to the two remaining transforms. Finally, the transformations will be compared and their usefulness in relation to image processing will be explained.
Algorithmische Komposition
(2018)
Algorithmic composition is an interdisciplinary topic that unites music and science. The computer is able to generate algorithmic music with the aid of a specific algorithm. In this bachelor thesis, algorithmic composition is realized with the biology-inspired algorithms called Lindenmayer-system and cellular automaton. In order to realize the compositions, several techniques are presented as well as implemented and evaluated. Those techniques map the generated data from the algorithms on a meaningful musical result.
The present thesis describes the development of an OpenGL-based tool visualizing cavities of proteins, which can be observed during a static docking simulation. The goal is to achieve knowledge about interactions between proteins and ligands based on information about distances between them. At first chemical basics, which motivate the topic and are important for understanding the topic and the used algorithms, are presented. Furthermore existing software, which deals with similar issues, is described. Next the prerequisites for the development of the program are presented and the tool is described in detail. Concluding the tool is evaluated concerning performance and usage and a summarizing conclusion is given. The program turns out as a helpful tool for current research and a good base for further and deeper research projects.
In this bachelor thesis a code for astrophysical self-gravitating fluid
simulation is developed. The code runs mainly on the GPU. Minimal
simplifications of the physical model and some parameters for accuracy
and tuning allow simulations to be performed at interactive framerates
on most modern consumer grade computers that feature a dedicated
graphics card. It is used to simulate the birth of stars from a turbulent
molecular cloud. Multiple features of star formation, like accretion
discs and fragmentation, can be observed in the simulation, even when
low particle counts are used.
Helicopters are crucial in today’s life. A vast amount of applications prove
their range, which are not coverable by other types of aircraft. But they are
very complex systems, both, technically and physically. This is one of the
reasons why pilot training for helicopters is quite challenging. In the last
two decades flight simulators became a supplementary instrument in the
educational process of pilots. With flight simulators it is possible to replay
uncommon or dangerous situations. In this thesis a simple flight simulator
for helicopters will be developed based on rigid body physics. The foundation is a simplified rotor model which omits complex fluid dynamics. This
helps to keep the implementation simple and illustrative as well as provide simulation rates at real-time. The modules are implemented within
the Unreal Engine in such way, that changing helicopter characteristics is
very easy.