Speaker
Description
The Data Reference Method (DRM) is a method developed at PTB for determining and correcting the non-uniformity of the responsivity of the individual detector elements of two-dimensional resolution cameras. The method is based on an intended horizontal and vertical displacement of the camera in front of a radiation source and the successive recording of several (at least three) offset images of the radiation source. DRM is a scene-based method and does not require any reference sources. As a result, the DRM provides both the non-uniformity of the IR camera system used and the spatial radiation distribution of the radiation source under observation. To determine the non-uniformity of the IR camera, relative information about the response behaviour of the individual detector elements in relation to a freely selectable reference detector element is used, which is derived from the shifted image recordings. However, if the ITS-90 traceable temperature of the radiation source observed by the reference detector element is known, the thermal imaging camera can be absolutely calibrated.
The DRM is a procedure that enables the calibration of thermographic cameras even on inhomogeneous radiation sources, so that the requirements on the calibration source used can be significantly reduced without impairing the resulting calibration quality. The main contribution of the DRM in the calibration process is the improved determination of the non-uniformity of the response behaviour of the thermographic camera over the observed field of view almost independently of the quality of the spatial temperature homogeneity of the applied calibration source.
While, in theory, the method is not introducing an additional calibration uncertainty, in reality there are some restrictions that impair the achievable results. The main practical sources of uncertainty are the temporal stability of the radiance of the source and the responsivity of the thermal imager, i.e. the drift, the positioning precision when taking the images, and the short-term stability (noise) of the measurement. However, under typical conditions in a thermometry lab the DRM significantly improves the calibration and with it the performance of any thermal imager.
In this work, we discuss how these main uncertainty contributions when performing the DRM can be detected, how they can be corrected and give practical tips when performing the DRM. Furthermore, we will compile an uncertainty budget for practical exemplary calibration conditions. As part of the European project EPM JRP “23IND11 ThermoSI - Thermometry with embedded SI traceability for industrial applications”, we will employ a scene-based correction method for non-uniformity even when there is a non-uniform scene and identify the limits when there are e.g., discontinous temperature changes (e.g. sharp edges) that cause artefacts. The results will be compared with an ex-situ non-uniformity correction (i.e. using a reference scene without discontinous temperature gradients in the lab). The preparatory work described here is fundamental for the successful implementation of the DRM approach in the project.
