Up to now, a large amount of computed tomography applications was in the field of flaw detection, failure analysis, dimensional measurements of geometrical features or statistical investigations of material properties. Often single slices were taken at well-defined places and used for further analysis. Recent developments in the hardware design and computational possibilities lead more and more to three-dimensional computed tomography, either directly by using a two-dimensional detector or indirectly by stacking together single slices. One of the reasons to do so is investigation of inner surfaces to measure dimensions.
For the moment the main application of such a technology is the first article inspection of cast components by comparing point cloud data with the CAD model. Deviations can be shown as colour maps on three-dimensional views or slices in any required orientation. Compared to traditional methods of first article inspection methods, the usage of a CT point cloud is almost faster and less expensive.
To verify the accuracy of this technology scanning parameters and surface segmentation methods have to be correlated with the generated point cloud. Test objects have been scanned and compared with co-ordinate measurement machine data. Adequate algorithms for surface segmentation have been developed at EMPA in co-operation with the Basle Institute of Technology. Instead of using two-dimensional-image processing methods an efficient three-dimensional-segmentation avoids discontinuous changes in the third direction. The resulting point cloud is an approximation of the isosurface of the objects in sub voxel accuracy.