Accurate Curvature Estimation along Digital Contours with Maximal Digital Circular Arcs

Abstract

We propose in this paper a new curvature estimator based on the set of maximal digital circular arcs. For strictly convex shapes with continuous curvature fields digitized on a grid of step h, we show that this estimator is mutligrid convergent if the discrete length of the maximal digital circular arcs grows in Ω(h^(-1/2)). We indeed observed this order of magnitude. Moreover, experiments showed that our estimator is at least as fast to compute as existing estimators and more accurate even at low resolution.

Jacques-Olivier Lachaud

Professor of Computer Science

My research interests include digital geometry, geometry processing, image analysis, variational models and discrete calculus.