Joint estimators for the specific intrinsic volumes of stationary random sets (Q2485860)

The authors consider stationary random closed subsets \(\Xi\) in \(\mathbb{R}^d\) that belong a.s. to the extended convex ring, i.e. \(\Xi\cap K\) can be represented as union of finitely many compact convex sets for any compact convex set \(K\). Applying the generalized version of Steiner's formula (for sets in the convex ring) to sets of the form \(E\Xi\oplus B_r(0)\) for balls \(B_r(0)\) with centre at the origin and radius \(r\) taking \(d- 1\) different values, say \(0< r_1<\cdots< r_{d-1}\), one gets a system of linear equations whose solution yields a joint estimator for the whole vector of the \(d- 1\) specific intrinsic volumes (among them the specific Euler-Poincaré characteristic and the specific surface area) of the underlying random set \(\Xi\). These unbiased nonparametric estimators can be represented as integrals over the stationary 0-1-valued random field associated with \(\Xi\). Moreover, conditions are established ensuring the mean-square consistency of these estimators.