拡張現実感のための複数仮説を用いたトラッキング手法(「VRにおける画像処理技術」特集)

概要

論文の詳細を見る
Many kinds of tracking for Augmented Reality had been proposed. In case of the feature point tracking, the pose is computed by minimizing the error between the observed 2D feature points and the back-projected feature points from the 3D scene model. This minimization problem is usually solved by a non-linear optimization. The main advantage of this approach is its accuracy. However, it is difficult to compute the correct pose unless an appropriate initial value is used. In addition, when some errors are included in the observation, this approach does not guarantee the correct pose even if it converged on the global minimum. So, once an incorrect pose was computed in one frame, tracking may fail in the next frame or the result will get farther from the correct one. In this paper, we propose a new tracking framework for augmented reality. Proposed method tracks features as multiple local hypotheses based on not only one pose but also multiple poses that are computed in the pose estimation in the previous frame. Since multiple poses are maintained as global hypotheses, as far as the correct pose is contained in the hypotheses, the tracking can be continued in even hard situations like a simple iterative scene with high-speed movement.
2007-06-30