As the one of new generations of human-machine interaction technologies, Augmented Reality (AR) is an important direction of virtual reality technology. With the computer-generated information, it can enhance the people’s perception to the real world and can be used in many fields. In order to align the virtual objects with real objects in the user’s view seamlessly, AR systems need accurately and quickly tracking the position and orientation of the camera. Thus, tracking is a very important issue in real-time AR systems. The resolution of tracking problem will promote the development and application of AR technology. This dissertation focuses on tracking problem in AR. The main content is as following: First, the problems arising from planar pattern-based tracking systems in AR are analyzed. By using the widely used AR toolkit-ARToolKit, we carry out the real experiments of augmented reality. Also simulation to the tracking in the AR systems is implemented in Matlab. The results of both two experiments reveal the existence of pose ambiguity problem and jitter problem of virtual object. An analysis to these problems is presented, which does some good to the development of planar-based AR systems. Second, based on the theory of automated inequality-type theorem proving and the relevant toolkit named “DISCOVERER”, P4P problem of 4 control point-based AR tracking systems is investigated. The PnP problem is briefly reviewed. To utilize the method of the inequality automated discovering, an equation system for P4P problem in AR is formulated. Moreover, a special case of P4P problem is computed and the complete distribution of real solutions is given. Based on these determinant conditions, the unstable points for AR systems can be avoided in practice, which can enhance the robustness and accuracy of AR systems. Third, various methods and strategies for tracking technology in marker-less AR systems are summarized. By comparing the different tracking methods, we choose the “Sift” invariant feature transform algorithm as the basis of tracker design. Furtherly taken advantage of the homography matrix, the pose of camera can be estimated. And this is our simple idea for the design of tracker. On the other hand, since edge and texture features have important roles in the tracking for AR systems, the denoising procedure in AR image preprocessing must be well-designed as to protect these valuable features. In this thesis, the PDE-based denoisin...
修改评论