Academic Projects

Master's Thesis Volume Renderer Bachelor Thesis

My academic projects were implemented as part of the visualization software Scifer of the Visual Information Analysis group at TU Kaiserslautern (formerly Computer Graphics and Visualization group at Heidelberg University). Scifer is written in C++ and based on OpenSceneGraph and Qt for visualization and interaction purposes.

Master's Thesis (2015)

Time slices

In my master's thesis I adressed time-dependent flow data. I developed and implemented a statistics based method to analyse time-dependent behaviour of vector fields. It was built upon data depth for pathlines, previously used in Mirzargar et al.'s work on a boxplot-like visualization of curves. By applying some simple adaptions I was able to make the computation of pathline data depth more robust, more precise and faster by approximately an order of magnitude. The method focuses on changes in behaviour. For a data set of wind velocities simulated for an entire year in six-hour-steps the method was able to find and highlight regions that coincide with known weather patterns.

If you want to learn more, you can find the thesis here—or feel free to contact me.

Volume Renderer (2014)

Time slices

The centrepiece of this Volume Renderer is its GLSL part. It exploits the nature of fragment shaders to efficiently process volume data.
In general, the fragment shader of an OpenGL pipeline is executed for each pixel on the screen. This leads to a large amount of similar tasks which are processed in parallel by GPU implementations. We can exploit this fact by implementing a ray-tracer in the fragment shader: For each pixel we sample along a ray that is "shot" into the scene—in our case the volume data set. The sampled values can then be processed further. Among the inevitable early ray termination this project features a flexible design that allows to easily swap out the value processing method. Along withe already implemented Maximum Intensity Projection, one- and two-dimensional transfer functions, the client can provide a method of his or her own as a shader function.

Bachelor Thesis (2013)

Extracted brain Extracted brain

My bachelor thesis was set in the realms of three-dimensional data. I implemented and analysed a method first presented by Ip et al. to segment volume data based on the two-dimensional gradient-intensity-histogram of the input data. A semi-automatic segmentation is obtained by transforming the mentioned histogram into a graph and calculating minimal normalized cuts. The segmentations can then be used, for example, to apply different colours to different materials using transfer functions.

The figure above shows two examples of such semi-automatically obtained visualizations and their respective histograms. As you can see, the method was able to isolate material that represents the brain.

The thesis (German) can be found here. It was written at the CoVis group of Heidelberg University under supervision of Christian Heine, Heike Leitte and Bastian Rieck.