Enhanced Quasi-reality-based Visual Tracking With Technology Research

Augmented Reality (AR) has become an important research field in VR and next generation of human computer interface. AR systems add virtual computer-generated material to the surrounding physical world to improve the person's perception. One of the most basic problems currently limiting AR applications is the registration problem. The objects in the real and virtual worlds must be properly aligned with respect to each other.This thesis foucus on the marker tracking and mark-less tracking.Firstly, we make a full analysis of L-K algorithm in image alignment. We compare and evaluate the inverse compositional algorithm and hyperplane approximation algorithm, and develop a matlab program to test their efficiency and convergence rates.Secondly, we design a new kind of fiducial marker which consists of a square outline with code region and visual picture in the interior. Compared with other fiducial marker systems, such as ARtoolkit, etc, this system is more usable and accurate. Based on the feature of the fiducial marker, we propose an algorithm for fiducial marker tracking. Robust tracking is obtained by combining corner amd texture information. Experiment result shows the marker system and the tracking algorithm are stable and effective.Thirdly, we research the Wide Baseline Matching of feature points and treat this problem as a classification problem. A tracking system based on arbitrary textured planar pattern is proposed. This system can easily be used for Augmented Reality applications.