Cooperative sensor data fusion for contactless quantitative 3D thermal imaging

Sebastian Schramm, M. Sc.

Since July 2017

Federal State of Hesse

Two-dimensional thermography is an established method for inspecting plants and objects. The advent of new and low-cost 3D sensor technology has already led to studies of first 3D thermal imaging systems. The user gains additional spatial information about the measured object, avoids projection-related measurement errors and also improves the visualization of energetic weak points. Such a system has also been developed at the Department of Measurement and Control at the University of Kassel (see figure). Previous international research work has focused on the creation of 3D thermograms from the sensor data fusion of the individual measurement channels, assuming the emissivity of the measured object to be known.

However, in order to be able to determine the surface temperature of a measured object from the emitted heat radiation, information about emissivity and possible reflections is required. For the estimation of emissivity in qualitative measurements, it is usually sufficient to use lookup tables with material-dependent standard values. For higher demands when using quantitative thermal measurements, adhesive strips or coatings with known high emissivity, boreholes and the comparison with contact temperature measurements are state-of-the-art methods. For these, the measuring object must be touched or even changed, which is practically not always possible or desirable. Although there is great interest in an user-friendly determination of emissivity, there is not much scientific work that addresses this issue.

Within the scope of the research project, non-contact and non-invasive methods for estimating emissions have to be investigated and characterized experimentally in the laboratory and by means of industrial case studies (e. g. thermal processing plants). Possible interference radiation must be reliably detected and excluded from modeling. Methods for segmenting the 3D models could also make it possible to examine different surfaces with varying properties at the same time.