Graphical modelling of measurement uncertainties in vision-based metrology

Measurement systems perform a quantitative comparison of an unknown physical quantity with a known reference. Vision sensors used in metrological applications provide a non-intrusive and non-invasive way to estimate geometric measurands and are, therefore, well suited for many industrial applications. In recent years the availability of high-resolution sensors and adequate processing power has led to an increased importance of vision-based measurement applications. This paper is concerned with the evaluation of measurement uncertainties in vision-based applications. In particular, we discuss the applicability of Gaussian uncertainties in vision-based metrological applications and present a frame-work for the uncertainty propagation of Gaussian quantities. The frame-work includes a guideline to model the measurement process based on the cause-effect diagram using simple graphical building blocks.