Evaluation of Reconstruction Methods in X-ray Computed Tomography Geometric Measurement

作     者：Kaojie Yue Huan Shao Stefano Petrò Giovanni Moroni 

作者机构：Department of Mechanical Engineering Politecnico di Milano Via La Masa 1 Milano 20156 Italy 

出 版 物：《Procedia CIRP》 

年 卷 期：2024年第129卷

页      面：139-144页

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 08[工学] 

主　　题：X-ray computed tomography Geometric measurement Reconstruction algorithm 

摘      要：In X-ray Computed Tomography (XCT) geometric measurement, the reconstruction process is a main step that significantly influences acquisition strategies and measurement results. Analytical algorithms, primarily Feldkamp-Davis-Kress (FDK), and algebraic algorithms are commonly used for reconstruction, with FDK being the dominant choice for dimensional measurements. Although algebraic algorithms offer advantages like denoising and handling limited projections, their evaluation concerning measurement accuracy remains insufficient. To assess the performance with respect to measurement accuracy and sensitivity to acquisition parameters, we performed typical reconstruction algorithms on the measurements of diameter, cylindricity, and flatness, including FDK and ordered subsets simultaneously algebraic reconstruction technique (OS-SART). Moreover, the algorithm comprised in a commercial software was used and evaluated. Variables in acquisition included part orientation and exposure time. The analysis comparing XCT measurements with those from Coordinate Measuring Machine showed different performances of the two reconstruction algorithms and the potential of algebraic algorithms in XCT geometric measurement.