University of Bergen | Faculty of Mathematics and Natural Sciences | Department of Informatics | Visualization Group
You are here: Department of Informatics > Visualization Group > Master Thesis > Åsmund Birkeland
 > about
 > team & contact info
 > research
 > publications
 > projects
 > teaching
 --- > Master Thesis
 > seminars
 > resources
 > network
 > events
 > links

View-Dependent Peel-Away Visualization for Volumetric Data

Master Degree Thesis
by Åsmund Birkeland (
supervised by Ivan Viola




Traditional illustration of three-dimensional structures, has developed techniques to provide clear view on internal features that are otherwise hidden underneath other outer structures. Many techniques attempt to create a better view on features of interest, while still conveying information about surrounding contextual structures. In illustrative volumetric visualization, techniques developed in early illustrations have been adopted to increase visibility and improve the comprehension of volumetric datasets during interactive visualizations.

In this thesis a novel approach for peel-away visualization is presented. Newly developed algorithm extends existing illustrative deformation approaches which are based on deformation templates and adds a new factor of view-dependency to the peeling process. View-dependent property guarantees the viewer unobstructed view to structure of interest. This is realized by rotating the peel-template so that the structures peeled-away always face away from the viewer. Furthermore the new algorithm computes the underlying peel-template on-the-fly, which allows animating the level of peeling. This is very helpful for understanding the original spatial arrangement. When structures of interest are tagged with segmentation masks, an automatic scaling and positioning of peel deformation templates allows guided navigation and clear view at structures in focus as well as feature-aligned peeling. The overall performance allows smooth interaction with reasonably sized datasets and peel templates as the implementation maximizes utilization of computation power of modern GPUs.


#1 #2 #3
 Last change: Ivan Viola, 2009-01-07