EE Seminar: Amit Boyarski
~~Electrical Engineering-Systems Department
*** SEMINAR ***
M.Sc. student under the supervision of Prof. Alex Bronstein
on the subject:
Optimization on Distance Maps with Applications to Deformable Shape Processing
Distance maps arise in a variety of engineering fields ranging from geometry processing to geophysical sciences. In these fields, it is common to encounter an optimization problem involving distances between points lying on a Riemannian manifold. The distance between sets of points on the manifold depends on both the source and the destination points, as well as on the metric of the manifold. These quantities can be used to encode geometric information about an object. In order to optimize over them, for example, using first order optimization methods, a fast computation of the sub-gradient is required. In this work, we propose a fast way to compute sub-gradients of the distance map with respect to the source points and the Riemannian metric tensor, and explore problems in geometry processing and medical imaging that require optimization on distance maps.