Predicting the Accuracy of Optical Tracking Systems

Optical tracking is a technology to localize the position and rotation of objects continuously with the help of optical cameras. Optical tracking is a basic technology used inmany applications. In augmented reality tracking is used to augment real objects with virtual informations. To place the informations at the right position, the position and viewing direction of the observer and that of the objects to be augmented have to be determined. To determine an accurate relationship between real and virtual object it is necessary to take the measurements as precise as possible, but even with great diligence and precision noisy errors will be inevitable. The noise on the image plane of the camera can lead to considerable errors in the localization of the objects. In certain applications for instance in themedical domain it is necessary to know andminimize these errors a priori and at runtime The noise in measurements can be described with a gaussian error model. Together with the error propagation rules it is for instance possible to predict the localization error. With the help of the gaussian error model and the error propagation rules several mathematical models are developed which allow to predict the expected uncertainty based on noise errors. The error propagation is widely used in this thesis eg. for the two developed applications which can quantitatively predict the accuracy of an optical tracking systemwith more than two cameras.