Measuring Similarity Between Curves on 2-Manifolds via Homotopy Area

TL;DR

This paper introduces a new, efficiently computable similarity measure for curves on 2-manifolds based on the minimal surface area swept during homotopy, applicable to planar and higher-genus surfaces.

Contribution

It defines a novel similarity measure for homotopic curves on 2-manifolds and provides efficient algorithms for computing it in planar and triangulated surfaces.

Findings

Algorithms run in quadratic or near-linear time.

The measure effectively captures curve deformability.

Applicable to curves on various 2-manifolds.

Abstract

Measuring the similarity of curves is a fundamental problem arising in many application fields. There has been considerable interest in several such measures, both in Euclidean space and in more general setting such as curves on Riemannian surfaces or curves in the plane minus a set of obstacles. However, so far, efficiently computable similarity measures for curves on general surfaces remain elusive. This paper aims at developing a natural curve similarity measure that can be easily extended and computed for curves on general orientable 2-manifolds. Specifically, we measure similarity between homotopic curves based on how hard it is to deform one curve into the other one continuously, and define this "hardness" as the minimum possible surface area swept by a homotopy between the curves. We consider cases where curves are embedded in the plane or on a triangulated orientable surface with genus $g$, and we present efficient algorithms (which are either quadratic or near linear time, depending on the setting) for both cases.