Analysis of reconstruction of functions with rough edges from discrete Radon data in $\mathbb R^2$

Abstract

We study the accuracy of reconstruction of a family of functions $f_ε(x)$, $x\in\mathbb R^2$, $ε\to0$, from their discrete Radon transform data sampled with step size $O(ε)$. For each $ε>0$ sufficiently small, the function $f_ε$ has a jump across a rough boundary $\mathcal S_ε$, which is modeled by an $O(ε)$-size perturbation of a smooth boundary $\mathcal S$. The function $H_0$, which describes the perturbation, is assumed to be of bounded variation. Let $f_ε^{\text{rec}}$ denote the reconstruction, which is computed by interpolating discrete data and substituting it into a continuous inversion formula. We prove that $(f_ε^{\text{rec}}-K_ε*f_ε)(x_0+ε\check x)=O(ε^{1/2}\ln(1/ε))$, where $x_0\in\mathcal S$ and $K_ε$ is an easily computable kernel.