Study On 3D Registration Algorithm For Augmented Reality

Augmented reality (AR) is a new hot research field, which is a combination of many subjects, as for the fact that 3D registration is one of the most key technologies for augmented reality, which attracts many universities, institutions and scholars engage in the research of this issue. As we know, the accuracy of the registration has a direct relation to an AR system, to some extent, which determines the result to be success or not of an AR system. Therefore, it's essential and pressing to pursue for the research of 3D registration for augmented reality.This paper expatiated the composition, the characteristics and the key technologies of an AR system based on the foundation of an exploration of the status quo for the development in augmented reality both home and abroad. Analyzed and compared several registration algorithms, then a detailed principle of the registration algorithm based on marker detection is given, and deduced the process of this algorithm. Particularly, analyzed the factors affect the accuracy of camera calibration, and ameliorated corresponding algorithm. To sum up, in order to improve the accuracy of 3D registration, this paper adopted a special method to tract and predestinates the marker, designed kalman filter to ensure the accuracy of circumgyration angle and to decrease tracking dimension.At the last part of this paper, after undergoing experiment of different markers, the result of the experiment shows that the number of marker corner, the light of experiment condition, the angle between the lens and the marker adopted affects the accuracy of registration. Beyond that, at the conclusion of this paper a total summarize is given, unsolved problems are analyzed, and the issue need to improve and perfect which related to the future work was also pointed out.