Für eine korrekte Darstellung dieser Seite benötigen Sie einen XHTML-standardkonformen Browser, der die Darstellung von CSS-Dateien zulässt.


Invited Talks

Prof. Dr. Daniel Cremers

Technical University Munich

Variational Methods and Convex Relaxation Techniques for Computer Vision

Numerous computer vision problems can be solved by variational methods and partial differential equations. In my presentation, I will show how problems like image segmentation, stereo and 3D reconstruction can be formulated as variational problems. Subsequently, I will introduce methods of convex relaxation and functional lifting which allow to compute globally optimal or near-optimal solutions. Experimental results demonstrate that these spatially continuous approaches provide numerous advantages over spatially discrete (graph cut) formulations, in particular they are easily parallelized (lower runtime), they require less memory (higher resolution) and they do not suffer from metrication errors (better accuracy).

Prof. Dr.-Ing. Marcus Magnor

Computer Graphics Lab
Technical University Braunschweig

Computer Graphics 2.0 - The Virtual World is not Enough

Expectations on computer graphics performance are rising continuously: whether in computer games, training simulators, or broadcast production, ever more realistic rendering results are to be achieved at real-time frame rates. In fact, thanks to progress in graphics hardware as well as rendering algorithms, visual realism is today within easy reach of off-the-shelf PCs.

With rapidly advancing rendering capabilities, the modeling process is becoming the limiting factor in computer graphics. Higher visual realism can be attained only by having available more detailed and accurate scene descriptions. So far, however, modeling 3D geometry and object texture, surface reflectance characteristics and scene illumination, motion and emotion is a labor-intensive, tedious process. The cost of visually authentic content creation using conventional approaches increasingly threatens to stall further progress in realistic rendering applications.

My talk centers on alternative modeling approaches. I will discuss and exemplify different methods on how to recover digital models of real-world entities. Suitable models may be derived based on the physics of the scene or by regarding primarily perceptional consequences. While the former approach yields physically meaningful information about the scene, approaches of the latter kind allow for easier modeling and natural-appearing rendering results. This opens up various new opportunities, extending the scope of computer graphics beyond virtual worlds to encompass visual reality.

Prof. Dr. Anders Ynnerman

Department of Science and Technology
Linköpings Universitet, Sweden

Rendering and Interacting with Large Scale Volumetric Data for Medical Applications

The talk will present recent advances in medical volume rendering from the Center for Medical Image Science (CMIV) and the Norrköping Visualization and Interaction Studio (NVIS), both at Linköping University in Sweden.

The first part of the talk will address the issue of data reduction and multi resolution representations for Level-of-Detail selection using the knowledge encoded in transfer functions. Knowledge encoding can also be used to obtain fuzzy classification of unsegmented data and it will be demonstrated how classification can be used to improve transfer function design and enhance features of interest. Other aspects of medical volume rendering such as the use of illumination models to enhance depth cues and to convey additional information from other co-registered sources will also be covered.

The second part of the talk will present methods for haptic (force feedback) interaction with volumetric data. New methods for the design and implementation of haptic modes for medical data will be presented and haptic feedback for time resolved volumes will be demonstrated. Throughout the presentation medical examples of volume rendering will be shown such as full body virtual autopsies using the presented methods.