摘要
提出了一种新的增强现实跟踪注册方法。采用跟踪-学习-检测TLD算法,将目标跟踪任务分成跟踪器、学习器和检测器,通过跟踪器和检测器并行对视频流中的图像帧进行跟踪、检测操作,并使用P-N学习算法进行学习和更新,综合这三部分从而确定稳定的目标位置;在所跟踪到目标位置的基础上,使用基于局部SURF算法对目标的特征点进行检测与匹配操作;根据所获得的特征点计算出三维注册参数,完成增强现实系统的跟踪注册。实验结果表明,所提出的方法有效地解决了传统跟踪算法目标跟踪不稳定、易丢失目标和目标检测算法耗时严重、准确度不高等问题,使增强现实系统更稳定、更具鲁棒性。
A novel method of tracking and registration for Augmented Reality (AR) was proposed. Using Tracking-Learning-Detection method based on TLD algorithm, the tracking task was deposed into three parts which were tracker, learning and detector respectively. The tracker tracked the targets in the frames of video streams and the detector detected the targets in every frame. The learning learned data from the detector and tracker and updated them. Stable outputs were determined by these portions together. The feature points were detected and matched by using SURF algorithm on the basis of the positions which were tracked. The parameters were worked out which were required by 3-D registration according to the feature points to realize the goal of AR tracking and registering. The experiments show that the proposed method solves the tracking instable, missing targets easily, costing much time, getting the low precision compared with the traditional methods. And it makes the AR system more stable and more robust.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2014年第9期2062-2067,2072,共7页
Journal of System Simulation