Morphological Non-Rigid Registration
The registration problems concerns the problem of finding a deformation from one image into another. In time sequences for example one might be interested in the deformation or motion field of two subsequent images. This plays an important role for time-sequence analysis in medical imaging. Another research interest is the non-rigid registration of multi-modal data (e.g. CT, MRT). We use multiscale, multilevel and multigrid methods to solve the computationally demanding problem with PDEs.
Poster: Morphological Registration
Poster: Gradient Flows
Research Site: Morphological Image Registration
Research Site: Image Matching with application in Medical Imaging
Level Set formulations for geometric flows
We are interested in implicit formulations for geometric flows (e.g. surface diffusion, Willmore flow), based on Finite Elements in space and a semi-implicit time stepping scheme.
Research Site: A level set formulation for Willmore flow
Segmentation with adaptive Level Set methods
Segmentation is a key step for the evaluation and detailed study in a wide range of computer vision applications. It concerns the recognition of certain kinds of objects in given digital images, e.g. the extraction of objects like tumors in medical images. We use level set methods for the flexible numerical handling of surface evolutions on adaptive grids. This becomes very important for the handling of ever increasing resolution capabilities of scanning devices.
Reseach Site: Level Set Segmentation on Adaptive Grids