Determination of compactly supported functions in shift-invariant space by single-angle Radon samples
Abstract
While traditionally the computerized tomography of a function $f\in L^{2}(\mathbb{R}^{2})$ depends on the samples of its Radon transform at multiple angles, the real-time imaging sometimes requires the reconstruction of $f$ by the samples of its Radon transform $\mathcal{R}_{\emph{\textbf{p}}}f$ at a single angle $\theta$, where $\emph{\textbf{p}}=(\cos\theta, \sin\theta)$ is the direction vector. This naturally leads to the question of identifying those functions that can be determined by their Radon samples at a single angle $\theta$. The shift-invariant space $V(\varphi, \mathbb{Z}^2)$ generated by $\varphi$ is a type of function space that has been widely considered in many fields including wavelet analysis and signal processing. In this paper we examine the single-angle reconstruction problem for compactly supported functions $f\in V(\varphi, \mathbb{Z}^2)$. The central issue for the problem is to identify the eligible $\emph{\textbf{p}}$ and sampling set $X_{\emph{\textbf{p}}}\subseteq \mathbb{R}$ such that $f$ can be determined by its single-angle Radon (w.r.t $\emph{\textbf{p}}$) samples at $X_{\emph{\textbf{p}}}$. For the general generator $\varphi$, we address the eligible $\emph{\textbf{p}}$ for the two cases: (1) $\varphi$ being nonvanishing ($\int_{\mathbb{R}^{2}}\varphi(\emph{\textbf{x}})d\emph{\textbf{x}}\neq0$) and (2) being vanishing ($\int_{\mathbb{R}^2}\varphi(\emph{\textbf{x}})d\emph{\textbf{x}}=0$). We prove that eligible $X_{\emph{\textbf{p}}}$ exists for general $\varphi$. In particular, $X_{\emph{\textbf{p}}}$ can be explicitly constructed if $\varphi\in C^{1}(\mathbb{R}^{2})$. The single-angle problem corresponding to the case that $\varphi$ being positive definite is addressed such that $X_{\emph{\textbf{p}}}$ can be constructed easily.
- Publication:
-
arXiv e-prints
- Pub Date:
- November 2022
- DOI:
- 10.48550/arXiv.2211.08693
- arXiv:
- arXiv:2211.08693
- Bibcode:
- 2022arXiv221108693L
- Keywords:
-
- Mathematics - Functional Analysis
- E-Print:
- Journal of Functional Analysis, 2023