Application Assessment of OS-SART Reconstruction Algorithm with Limited Number of Projections in XCT Geometric Measurement

In industrial X-ray Computed Tomography (XCT), two primary kinds of reconstruction algorithms are utilized: the Feldkamp-Davis-Kress (FDK) algorithm and the algebraic reconstruction technique (ART) algorithms. While the FDK algorithm is predominantly employed in industrial XCT due to its competitive efficiency, it necessitates a large number of projections (Np) to yield good 3D images. Conversely, although ART algorithms require relatively longer computational times, their mechanism enables volume reconstruction with a smaller number of projections, thereby offering a potential advantage in scenarios with limited Np. To rigorously investigate the potential of ART algorithms under conditions of limited Np, this study focuses on the ordered subsets simultaneously algebraic reconstruction technique (OS-SART) algorithm, a classic ART algorithm. A series of experiments is conducted, targeting three key algorithmic parameters. These experiments are based on a stepped cylinder part across varying Np levels, with the error between the values of XCT measurement and coordinate measurement machine as the evaluation metric. Eventually, the impacts of these parameters are evaluated based on the experimental results.