The Hurst parameter of a fractional Brownian motion (fBm) on a smooth manifold determines the regularity of the corresponding fractional Brownian surface (fBs). Estimating the regularity of a given fBs is difficult since the underlying fBm is unknown. We propose here the first spectral-regression algorithm to estimate the Hurst parameter of a given fBs. The algorithm is evaluated on a set of simulated fractional Brownian spheres and we present an application to brain surfaces.

This presentation is part of Contributed Presentation “CP2 - Contributed session 2”

Authors:

Hamed Rabiei (Aix-Marseille University)

Frédéric Richard (Aix-Marseille University)

Julien Lefèvre (Aix-Marseille University)

Olivier Coulon (Aix-Marseille University, Institut de Neurosciences de La Timone)

Keywords:

brain surface complexity, fractional brownian motions on manifolds, hurst parameter estimation, spectral analysis, statistical inverse estimation methods