A stochastic analysis of the calibration problem for Augmented Reality systems with see-through head-mounted displays

摘要

This paper presents a closed stochastic solution for the calibration of see-through head-mounted displays (STHMD) for Augmented Reality. An Augmented Reality system (ARS) is based on several components that are affected by stochastic and random errors. One important component is the tracking system. The flock of birds (FOB) tracking system was tested for consistency in position and orientation outputs by establishing constraints that the system was required to meet. The tests for position and orientation were separated to derive uncorrelated quality measures. The tests are self-controlling and do not require any other measuring device. In addition, the image coordinate accuracy also had to be determined to complete the stochastic description of the calibration problem. Based on this stochastic model, different mathematical models were tested to determine whether or not they fit the stochastic model. An overview of different calibration approaches for optical see-through displays is given and a quantitative comparison of the different models is made based on the derived accuracy information.