Graphical models and simulation for THz-imaging

The utilization of terahertz (THz) radiation is an active research field. Approaches for non-intrusive material monitoring or reliable surveillance of hidden treats promise considerable improvements of current technologies in these research fields. Due to the prototypical nature of current methods, easily accessible or unoptimized techniques are commonly used for processing THz data or simulating THz radiation. Methods of computer graphics (CG) allow an efficient prototyping to evaluate the potential of these new THz technologies and algorithms.
In this thesis, CG methods are developed and applied to the problems of THz research. It is shown that these methods are beneficial in terms of efficiency and performance. A sparse voxel tree (SVT) is proposed to store highly detailed objects with inner structures for simulating the imaging capabilities of THz setups. Mainly, an efficient creation and rendering of this voxel structure are introduced. The SVT is applied to the simulation of a prototypical THz setup which scans a scene by focused radiation.
Another THz setup, which is based on unfocused radiation and synthetic aperture techniques, is used to show the benefits of GPU algorithms for the reconstruction of large THz data. Furthermore, a volume based simulation of this system is shown. In addition, a geometrical configuration of the THz system and the rendering of multimodal data from the THz setup are introduced.