Feasibility of Simulated Annealing Tomography

TL;DR

This paper introduces a novel approach to significantly accelerate simulated annealing tomography by using pre-calculated intersection lengths, making it more feasible for practical applications.

Contribution

The paper proposes a PILA-based method to reduce computational costs and incorporates a BoME strategy to enhance convergence speed in SAT.

Findings

Speed-up of over 300 times in SAT reconstruction

Improved convergence with BoME strategy

Effective performance demonstrated through numerical experiments

Abstract

Simulated annealing tomography (SAT) is a simple iterative image reconstruction technique which can yield a superior reconstruction compared with filtered back-projection (FBP). However, the very high computational cost of iteratively calculating discrete Radon transform (DRT) has limited the feasibility of this technique. In this paper, we propose an approach based on the pre-calculated intersection lengths array (PILA) which helps to remove the step of computing DRT in the simulated annealing procedure and speed up SAT by over 300 times. The enhancement of convergence speed of the reconstruction process using the best of multiple-estimate (BoME) strategy is introduced. The performance of SAT under different conditions and in comparison with other methods is demonstrated by numerical experiments.