Discrete Terahertz tomography: a simulation study

Abstract

As Terahertz (THz) technology becomes more prominent in the imaging industry, there is a rising need for improved reconstruction techniques for THz tomographic imaging. Conventional reconstruction techniques yield artifacts for limited data problems, hindering further analysis of the reconstruction. In this paper, we explore a discrete reconstruction algorithm for Terahertz tomography, which exploits prior knowledge on the gray levels in the imaging data of the object to obtain high quality reconstructions. Simulations show that, compared to THz-SIRT reconstruction, the discrete THz algorithm (THz-DART) generates more accurate reconstructions, especially when only a small number of projections or projections acquired within a limited angular range are available.