Invited Talk January 2007

Talk by Mikael Rousson

Mikael Rousson (Siemens Corporate Research (SCR), Princeton, USA) will be giving an invited talk on Wednesday, Jan 31, 16:00-18:00, at Klinikum Innenstadt, Seminarraum 1 (room nr 102)

on

For general information regarding the talk (such as how to get there), please contact Darko Zikic.

Abstract

Since its introduction as a means of front propagation and its first application to edge-based segmentation in the early 90's, the level set method has become increasingly popular as a general framework for image segmentation. In this talk, I will present several contributions to this framework.

• Statistical shape models: Given a set of training shapes, we perform a kernel density estimation in the low dimensional subspace spanned by the training shapes. In this way, we are able to combine an accurate model of the statistical shape distribution with efficient optimization in a finite-dimensional subspace. (Rousson and Cremers - MICCAI 2005).
• Efficient piecewise smooth intensity model: We propose a fast and robust segmentation model for piecewise smooth images. Rather than modeling each region with global statistics, we introduce local statistics in an energy formulation. The shape gradient of this new functional gives a contour evolution controlled by local averaging of image intensities inside and outside the contour. (Piovano and Rousson – Scale-Space and Variational Methods 2007).
• Interactive segmentation: We propose a probabilistic framework for image segmentation which integrates input intensity information and user interaction on equal footings. The resulting algorithm determines the most likely segmentation given the input image and the ser input. (Cremers, Fluck, Rousson and Shmuel – SPIE 2007).
• Introducing ordered spatial dependency for coupled segmentation: We propose a general framework to segment several structures. ather than inventing yet another segmentation algorithm, we introduce inter-structure spatial dependencies to work with existing egmentation algorithms. Ranking the structures according to their dependencies, we end up with a hierarchical approach that improves each individual segmentation and provides automatic initializations. (Rousson and Xu – MICCAI 2006).