A tracking and registration method based on ORB and KLT for augmented reality system

Registration is one of the key technologies of augmented reality system. In order to realize accurate and realtime registration and keep the right position and orientation of a virtual object in real world scenarios, a new registration method based on ORB and KLT algorithms is proposed. It detects the target from the video frame first and calculates the homographic matrix of initial camera poses, then continues to track camera pose for further registration. Error correction method and verification strategy are presented to improve the reliability and accuracy of the registration during the tracking process. Experiments show that the proposed method gains better registration accuracy and efficiency, it achieves real time combination of the virtual information with the moving targets.