Key points are not available for this paper at this time.
A mobile augmented reality (AR) game application was developed based on simultaneous localization and mapping (SLAM). The SLAM-based AR game requires to estimate the pose from the camera input image in real time. Before running the game, point cloud data for a real-world game space is built. While the game is running, the camera pose is estimated by matching the prebuilt point cloud data and camera input image. To minimize errors in the matching, we present a hybrid method using an optical flow and ORB descriptor, where the optical flow accurately tracks the displacement of keypoints in consecutive images, and the ORB is a fast keypoint descriptor under a BSD license. The performance of the hybrid method is compared with a method using only the ORB descriptor matching. In addition, a mobile AR game embedding the hybrid method was tested in both indoor and outdoor environments.
Bang et al. (Wed,) studied this question.