Bundle Adjustment Rules

In this paper we investigate the status of bundle adjustment as a component of a real-time camera tracking system and show that with current computing hardware a significant amount of bundle adjustment can be performed every time a new frame is added, even under stringent real-time constraints. We also show, by quantifying the failure rate over long video sequences, that the bundle adjustment is able to significantly decrease the rate of gross failures in the camera tracking. Thus, bundle adjustment does not only bring accuracy improvements. The accuracy improvements also suppress error buildup in a way that is crucial for the performance of the camera tracker. Our experimental study is performed in the setting of tracking the trajectory a calibrated camera moving in 3D for various types of motion, showing that bundle adjustment should be considered an important component for a state-of-the-art real-time camera tracking system.