Pore detection by x-ray computed tomography (CT) is widely used for additively manufactured (AM) parts. The ability of a CT device to reliably detect pores of a certain shape and size is still being investigated, though. Typical methods to evaluate the error of CT-based pore detection utilize reference objects or microscope images of cross-sections. This work improves the cross-section method by directly matching the two-dimensional (2D) plane depicted in the microscope image to the corresponding plane within the three-dimensional (3D) scan. For this purpose, a specially designed test part was scanned for a first time before preparing a cross-section and a second time afterward. The test part's design allowed to identify the cross-section's plane parameters relative to the part's local coordinate system in the second scan. The plane parameters were then transferred to the first CT scan in which pores had been detected to identify voxels that lay within the plane that was to become the cross-section. An image was extracted from this plane that directly matched a microscope image of the cross-section. Comparing CT and microscope images highlighted that CT's relative lack of resolution may lead to relevant errors when analyzing pore area and pore form.
Read full abstract