Current Research Activities

My PhD thesis is mostly concerned with motion estimation and parameter estimation of dynamically deformable surfaces in medical computer vision applications.

Basically, what I'm trying to do is helping a surgeon (who is doing a minally invasive surgery using an endoscope) by gathering information from video data, performing a 3D reconstruction from those video images, fitting the reconstructed data to a surface model, and finally estimating the motion parameters that explain the observed surface motion and deformation, also allowing motion prediction to a limited degree.

Based on these requirements, I am interested in the following topics:

• Surface reconstruction from video data using B-spline models, NURBS etc.
• Approximative physical models for deformable surfaces, e.g., mass-spring-models, chain mail algorithms etc.
• Parameter estimation techniques, e.g., Kalman filtering, particle filters
• Optimization algorithms

I am also working on a CoTeSys project in cooperation with Prof. Dr. Nassir Navab. The project is concerned with taskflow recognition of everyday manipulation tasks. In this project, I am working on finding and using features from multiple view videos.

Past research activities

Before I started working in medical computer vision for Prof. Burschka, I was focusing on the theoretical aspects of computer science, especially complexity theory and formal languages under supervision of Prof. Dr. Markus Holzer. My diploma thesis was concerned with the complexity of pencil puzzles, and applicability of SAT solvers to those puzzles.

In the scope of that work, I have proven the NP-completeness of the Heyawake pencil puzzle. This proof [4] was the topic of the first paper I have written in my academic career. It has been accepted at the FUN conference in 2007, and was subsequently published in the Springer LNCS series. The image to the right shows a Heyawake puzzle together with it's solution. For more information on this puzzle, see Wikipedia.

In addition to that, I have shown how SAT solvers can be used to check several kinds of pencil puzzles for unique solvability. The resulting software was very fast, yet easy to implement. It has been used as a basis for puzzle generating software at the company Axel Ruepp Rätselservice.

Publications

[1]	Dominik Steinhauser, Oliver Ruepp, and Darius Burschka. Motion segmentation and scene classification from 3D LIDAR data. Intelligent Vehicles Symposium, 2008 IEEE, pages 398-403, June 2008.
[2]	Oliver Ruepp and Darius Burschka. A geometrically inspired approach to active view planning. 2008. (http )
[3]	Hermann Gruber, Markus Holzer, and Oliver Ruepp. Sorting the slow way: An analysis of perversely awful randomized sorting algorithms. In Pierluigi Crescenzi, Giuseppe Prencipe, and Geppino Pucci, editors, FUN, volume 4475 of Lecture Notes in Computer Science, pages 183-197. Springer, 2007. (http )
[4]	Markus Holzer and Oliver Ruepp. The troubles of interior design-a complexity analysis of the game heyawake. In Pierluigi Crescenzi, Giuseppe Prencipe, and Geppino Pucci, editors, FUN, volume 4475 of Lecture Notes in Computer Science, pages 198-212. Springer, 2007. (http )