Deep learning has become a state-of-the-art method in machine learning, with a broad range of successful applications. Our goal is to explore the benefits of deep learning techniques for computational optical form measurements. The research is based on solving a nonlinear inverse problem aimed at the reconstruction of optical topographies from given processed interferograms. A U-Net network structure is chosen and tested on a simulated database. The obtained results are promising.

In this article we have investigated how the arbitrary orientation of a passive localization object in a magnetic field affects that field. For that we have identified the two sources of uncertainty, the geometrical extent of the object and the field inhomogeneity, and have quantified their effect analytically. Based on that we have compared the performance of several localization algorithms on that field variation. Results show that a dipole model is sufficient for the localization algorithm.

Computed tomography measurements can be subject to specific image artefacts, which can be dependent on the effective rotation axis of the work piece during the scan. The presented approach is to combine several CT scans with different rotation axes of the work piece using a data fusion approach. To improve the fidelity of the result, surface points are weighted individually within the algorithm, dependent on the local surface quality of the measurement.

Metrology has a key position in networked, adaptive production, with the task of a holistic and valid assessment of the state of various production scenarios. Current trends and challenges for metrology in networked production, e.g., multi-sensor systems, model-based measurements, virtual measurement processes or the integration into adaptable production systems are described and the concept of sensor information as a service is presented as a key development for the flexible use of metrology.

The analysis of aeroacoustic phenomena is crucial for deeper understanding of the damping mechanisms of a sound-absorbing bias flow liner. Simultaneous three-component velocity measurements of the superposed sound field and the flow field in a 3-D region of interest with over 4000 measurement points are presented. The natural Helmholtz–Hodge decomposition is applied to separate both fields from the measured velocity field in the spatial domain. This reveals new insight into the aerodynamic flow.

Deflectometric measurements of specular surfaces when performed with two reference planes lead to both data for surface points and surface normals. The deviations in the surface points are usually several orders of magnitude larger than those for the surface normals. In this paper we propose a method to fuse these data to increase the accuracy of the surface points. The results show that the accuracy can be increased by more than an order of magnitude.