Improved orthogonal iterative algorithm for pose estimation in augmented reality

Accurate pose estimation is significant in AR (Augmented Reality) system. Therefore, the registration algorithm should be fast and accurate. Most of the traditional algorithms cannot meet the demand. At present, the orthogonal iterative algorithm (OI) presented by Chien-Ping Lu is one of the best methods. However it requires a proper initialization. The research presents a new method based on OI. In which 4 coplanar markers are used to calculate the pose of camera in the initialization step. Furthermore, according to the character of the relative position of camera and markers, enantiomorphous problem introduced by coplanar markers is solved. It is proved via the experiments that the new method is steady, fast and appropriately calculates the pose of camera.