Real-Time Approaches for the Model-Based Reconstruction and Visualization of Flow Fields
In this thesis, three novel realtime methods for flow field exploration have been developed. First, a reconstruction method, which uses a physics-based filtering to generate a velocity field consistent with the underlying Navier-Stokes equations. Second, a focus for particle tracing (anchor lines), to analyze local flow features within the global flow context by visualizing how much particles separate over time. The benefit of using the finite time Lyapunov exponent for their seeding has been shown. Third, a method for diffusion tensor visulization based on GPU particle tracing. Interactivity of all methods is achived by using GPUs for computations and visulization. The efficiency of all methods has been shown on synthetic an real-world data sets.