Reconstruction of polygonal shapes from sparse Fourier samples (Q896804)

In this interesting paper, the authors reconstruct the characteristic function \(f(x_1,x_2) = 1_D(x_1,x_2)\) of a simply-connected polygonal domain \(D \subset {\mathbb R}^2\) from relatively few samples of the Fourier transform \(\hat f\). This reconstruction method is based on a stable Prony method (such as approximate Prony method, MUSIC or ESPRIT) for the recovery of univariate exponential sums. By this approach, the authors reconstruct the vertices of the polygon in a correct way. It is remarkable that this method works also for a non-convex polygonal domain \(D\). Note that the reconstruction of a convex polygonal domain \(D \subset \mathbb C\) from given moments were presented by \textit{G. H.~Golub} et al. [SIAM J. Sci. Comput. 21, No. 4, 1222--1243 (1999; Zbl 0956.65030)].