Application of methods of machine learning in CT measurement technology

X-ray computer tomography (R-CT) has already gained a well-established place among non-destructive testing methods for the inspection of product quality. Due to the detailed, high-resolution imaging of relevant component regions, the R-CT has enormous potential for solving current and future inspection tasks. On the other hand, there are often higher costs in comparison to other testing methods, which result from the investment in equipment, the scanning time as well as the subsequent analysis and evaluation time. As a service provider in the field of R-CT, Microvista GmbH specializes in the development of automated analysis and evaluation routines for CT layer images. The use of computer vision methods, especially for qualitative evaluations, which are usually carried out by an inspector, is not only seen as an essential step towards reducing inspection costs, but also to make the evaluation process more objective and thus more reliable. In the following, the associated challenges, in particular the handling of only a few training examples, as well as the results achieved will be presented using a practical example - fault detection in a water jacket.