Evaluating the standard uncertainty due to the voxel size in dimensional computed tomography

Abstract

X-ray computed tomography (XCT) is a promising tool for making dimensional measurements of complex engineering components. The adoption of XCT as a measurement tool is hindered by the inability to evaluate the uncertainty of XCT-based dimensional measurements; simply put, XCT users cannot specify how good (or bad) their measurements are. In this work, equations and a method are given to evaluate the standard uncertainty due to the voxel size; this being one of several sources of uncertainty in XCT-based dimensional measurements. It is envisioned that this standard uncertainty component will be combined with other standard uncertainties in a task-specific uncertainty budget, thus providing end users with a statement of XCT measurement uncertainty. It is claimed here that evaluating the standard uncertainty of the voxel size by means of a calibrated length leads to a traceable voxel size. For the example considered in this work, the voxel size is evaluated to be 80.005 μm ± 0.001 μm, with the voxel size uncertainty expressed as one standard deviation. When the standard uncertainty of the voxel size is propagated through to the final measurement result of a machined aluminium length bar, the standard uncertainty due to the voxel size is evaluated as ± 1.01 μm for a bi-directional length of nominally 55 mm.