Weight metamorphosis of varifolds and the LDDMM-Fisher-Rao metric
Hsi-Wei Hsieh, Nicolas Charon

TL;DR
This paper develops a metamorphosis framework for varifolds, extending geometric measure registration by combining diffeomorphic deformations with weight transformations, and introduces a Riemannian metric with practical computation methods.
Contribution
It introduces a novel metamorphosis model for varifolds with a new Riemannian metric and optimal control formulations, enhancing geometric measure registration techniques.
Findings
Well-defined Riemannian metric on varifolds with geodesics
Existence of solutions for the optimal control problems
Numerical methods for computing metamorphoses between varifolds
Abstract
This paper introduces and studies a metamorphosis framework for geometric measures known as varifolds, which extends the diffeomorphic registration model for objects such as curves, surfaces and measures by complementing diffeomorphic deformations with a transformation process on the varifold weights. We consider two classes of cost functionals to penalize those combined transformations, in particular the LDDMM-Fisher-Rao energy which, as we show, leads to a well-defined Riemannian metric on the space of varifolds with existence of corresponding geodesics. We further introduce relaxed formulations of the respective optimal control problems, study their well-posedness and derive optimality conditions for the solutions. From these, we propose a numerical approach to compute optimal metamorphoses between discrete varifolds and illustrate the interest of this model in the situation of…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
Topics3D Shape Modeling and Analysis · Morphological variations and asymmetry · Geometric Analysis and Curvature Flows
