Year: 2026
Type: preprint
Abstract: In Limited-View Computed Tomography (LVCT), one restricts the number of projection angles (or views) for data reduction, which has many implications, especially in medical domain. In most of the existing works on LVCT, the views are sampled randomly or in equi-spaced manner, which makes the sampling process oblivion to the image reconstruction solvers. Motivated by this gap between the sampling strategy and the image reconstruction solvers, we propose a deterministic way of finding an optimal set of views for LVCT using block-sparse optimization problem. The proposed methodology is tied to a selection of Radon sub-matrix possessing low total coherence which is desired for sparsity-based CT image reconstruction. We demonstrate efficacy of the proposed deterministic LVCT in recovering CT images.
Source: Research Square
Authors: Munnu Sonkar, Pradip Sasmal, CS Sastry
Language: English
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7235401057
