A Compressed Sensing-Based Algorithm and Simplified System to Improve the Efficiency of CW THz CT Imaging

Terahertz computed tomography (THz CT) represents a promising nondestructive testing (NDT) modality; however, the limited imaging efficiency of current THz CT systems and the quality of reconstruction algorithms restrict its widespread application. This study proposes a compact THz CT system, along with a CT reconstruction algorithm capable of producing accurate images from a small amount of data. Specifically, an ordered subsets expectation maximization-total variation minimization (OSEM-TV) algorithm based on compressed sensing theory is introduced, which substantially reduces artifacts and noise in reconstruction results under sparse projection angles, achieving high-quality reconstructions using only a small amount of projection data. Additionally, by utilizing a vector network analyzer (VNA) and minimizing the use of mirrors, the imaging system has been optimized for greater integration. Both simulation and experimental results validate the proposed algorithm’s advantages, particularly in performing 3D visualization of a pig hock bone under sparse projection angles, thereby extending the applicability of THz CT to biological materials.