Adaptive surface geometry determination in multi-material x-ray computed tomography using fringe projection

One of the main challenges during digital post-processing of x-ray computed tomography (XCT) measurement data is the reconstruction of the surface geometry of the measured objects. Conventionally, the surface geometry is defined as an isosurface with identical greyscale values or via gradient-based surface geometry determination. However, these approaches are susceptible to measurement artefacts. Due to the complex surface geometry and rough surface, XCT measurements of additively manufactured (AM) parts are particularly prone to measurement artefacts caused by various physical effects when the x-rays penetrate the material. The irregular greyscale values at the measured surface geometry render commonly used single threshold greyscale value based isosurfaces as insufficient for representing the external and internal surface of the measured objects. This issue becomes particularly apparent when measuring multi-material objects, such as additively manufactured polymer objects with integrated radio-frequency identification tags. To address this challenge, this study presents a methodology for reliable surface geometry determination of XCT data based on previously acquired fringe projection (FP) data. For this purpose, the conventionally acquired surfaces geometries from the XCT and FP measurements are extracted, pre-processed and registered to each other before being merged into a single mesh. This merged data set is subsequently used as a starting point or reference for a locally adaptive threshold surface geometry determination algorithm, which is able to capture the surface geometry at a sub-voxel resolution. In order to validate the methodology and confirm the envisaged benefits, selected geometry elements of the resulting surface geometry from measurements samples manufactured by additive manufacturing with integrated RIFD tags are compared to coordinate measurement machine reference measurements. The results indicate a more robust surface geometry detection against artefacts especially for multi-material applications where the x-ray absorbance of the inner material is higher than the absorbance of the outer material. In summary, the main result is a systematic approach to merge FP and XCT data, in order to combine the advantages of both techniques, which leads to an artefact-reduced surface geometry determination for multi-material applications and a reduction of deviations from reference measurements compared to conventional methods by nearly 35%.