Comparative study of marker-based camera tracking using extended and unscented Kalman filters

Camera pose tracking is an important issue for several modern applications like augmented reality, surveillance systems, robot localization, etc. In this paper, we compare the performances of unscented and extended Kalman filtering for improving camera pose estimation. For that, we propose to use external square markers to provide an accurate motion measurements. Several experiments are achieved in order to compare the accuracy of the two filters. Our results show that, for this kind of applications, the extended Kalman filter (EKF) can perform better than an unscented Kalman filter (UKF) but at a much lower computational cost.