Transportation Distances on the Circle and Applications

TL;DR

This paper studies Monge-Kantorovich distances for circular distributions, providing efficient computation methods and exploring applications in image feature matching and color analysis.

Contribution

It proves a method to compute distances on the circle by cutting it into a line, enabling linear-time calculations for circular histograms and applications.

Findings

Distance computation reduces to a linear problem after cutting the circle.

Linear-time algorithm for circular histogram comparison.

Applications in image feature matching and color hue analysis.

Abstract

This paper is devoted to the study of the Monge-Kantorovich theory of optimal mass transport and its applications, in the special case of one-dimensional and circular distributions. More precisely, we study the Monge-Kantorovich distances between discrete sets of points on the unit circle, in the case where the ground distance between two points x and y is defined as h(d(x,y)), where d is the geodesic distance on the circle and h a convex and increasing function. We first prove that computing a Monge-Kantorovich distance between two given sets of pairwise different points boils down to cut the circle at a well chosen point and to compute the same distance on the real line. This result is then used to obtain a metric between 1D and circular discrete histograms, which can be computed in linear time. A particular case of this formula has already been used in [Rabin, Delon and Gousseau SIAM 09}] for the matching of local features between images, involving circular histograms of gradient orientations. In this paper, other applications are investigated, in particular dealing with the hue component of color images. In a last part, a study is conducted to compare the advantages and drawbacks of transportation distances relying on convex or concave cost functions, and of the classical L-1 distance.