Modulbeschreibung

The students will learn how to reconstruct the three-dimensional world and the camera motion from multiple images. For this, focus lies on the determining correspondences between points in various images and respective constraints that allow to compute motion and 3D structure. A particular emphasis lies on mathematical descriptions of rigid body motion and of perspective projection.
For estimating camera motion and 3D geometry both spectral methods and methods of nonlinear optimization are used.

Multiple View Geometry is the theory of determining geometic properties of objects from one or more source images. These techniques are widely used in robotics and 3D scene recognition to produce meshes.
In order to improve the detection of feature points from stereoscopic source images, new methods should be implemented that make use of multiple view geometry. The assumption should be tested, that this will increase the amount of detected points.