Interactive Exploration of 3D Point-Clouds
This PhD thesis deals with the interactive exploration and manipulation of a generated 3D environment, for example large indoor spaces or cultural heritage sites, in an immersive virtual reality setup. For this, we use a previously captured 3D point cloud of the site as input data.
The candidate will investigate algorithms to process very large 3D point clouds in order to allow high-fidelity display both for the geometry as well as for the color channel (see Figure 6.10.1). The candidate will further investigate new data structures and algorithms as well as new interaction metaphors for 3D selection, manipulation, and interactive segmentation of interesting parts within the point cloud (see Figure 6.10.2). The algorithms should be designed for a range of immersive output options, ranging from a desktop 3D stereo setup, over a high-quality tourist installation, to a fully immersive setup using our VR framework in combination with a head mounted display. By utilizing the upcoming omnidirectional treadmill Virtualizer (https://www.ims.tuwien.ac.at/projects/virtualizer, see Figure 6.10.3), to which we currently have exclusive access, the observer can travel through the virtual world using natural walking.
A suitable background in computer graphics (in particular real-time rendering and GPU programming) and virtual reality is required. In addition, a strong mathematical background, e.g., optimization methods, is advantageous.
This topic is supervised by a team of 3 supervisors. Lead supervisor is Michael Wimmer (Institute of Computer Graphics and Algorithms). Additional supervisors are Annette Mossel (Institute of Software Technology and Interactive Systems) and Ulrike Herbig (Institute of History of Art, Building Archaeology and Restoration).
Figure 6.10.1: Example of high-quality reconstruction for geometry and color: Left: colored input point cloud Right: high-quality reconstruction with optimal color mapping.