Course: Correspondence Problems in Computer Vision (6622-611)

Persons:

• Prof. Dr.-Ing. Andrés Bruhn (verantwortlich)

Type of Course:

lecture with exercise

In-Class Hours Per Week:

4

Contents:

- Basic Approaches: Block Matching, Occlusion Detection, Interest Points, Feature Matching
- Optic Flow: Local and Global Differential Methods, Parametrisation Models, Constancy Assumptions, Data and Smootness Terms, Numerics, Large Displacements, High Accuracy Methods
- Sterep Matching: Projective Geometry, Epipolar Geometry, Estimation of the Fundamental Matrix
- Scene Flow: Joint Estimation of Structure, Motion, and Geometry
- Medical Image Registration: Mutual Informaion, Elastic and Curvature-Based Regularisation, Landmarks
- Particle Image Velocimetry: Div-Curl-Regularisation, Incompressible Navier Stokes Prior

Literature:

O. Faugeras, Q.-T. Luong: The Geometry of Multiple Images, 2001

J. Modersitzki: Numerical Methods for Image Registration, 2003

A. Bruhn: Variational Optic Flow Computation: Accurate Modeling and Efficient Numerics, Ph.D. Thesis, 2006

Location:

Stuttgart-Stadt

Module:

• 6622-610 Correspondence Problems in Computer Vision (compulsory)