Approximation of the emission coefficient for thermography by the combination of geometrical and spectral information -- ThermoHead

Abstract: Thermography allows for the remote measurement of surface temperatures and is widely used for the identification of energy losses, damage detection or quality control. However, thermal imaging is strongly material dependent and therefore measured and real temperatures can differ significantly. The emission coefficient resides between 0 and 1 and changes mainly with the electrical conductivity of the surface. While non-conductive surfaces (stone, wood, glass etc.) have a high emissivity of ca. 0.9, metallic surfaces reside around 0.2. If e.g. an object with a temperature of 100$^\circ$C and E=0.9 gets recorded via thermography, the recording shows a temperature of 90$^\circ$C. The same conditions for a metallic surface on the other hand can result in a measured temperature of only 20$^\circ$C, so room temperature. This article introduces a novel method how to approximate the electrical conductivity of surfaces by the combination of geometric and spectral information. A new thermographic panorama system called "ThermoHead" is used to generate high-resolution and geometrically calibrated thermal images of the surrounding. For the geometric information, a 3D point cloud from the very same point of view will be recorded with an arbitrary commercial available terrestrial 3D laser scanner. Since both data sets have been recorded successively from the exact same point of view, the congruent combination allows to transfer all geometric informations from the terrestrial 3D laser scanner to the thermographic measurement. The radiometric intensity depends on e.g. the angle of incidence of the thermographic measurement, which is now known and can be used to narrow down the electrical conductivity of the measured object.