Effects of Optical Combiner and IPD Change for Convergence on Near-Field Depth Perception in an Optical See-Through HMD

Effects of Optical Combiner and IPD Change for Convergence on Near-Field Depth Perception in an Optical See-Through HMD
复制标题

DOI:
10.1109/tvcg.2015.2440272
复制
发表时间:
2016-05
影响因子:
5.2
通讯作者:
Sangyoon Lee;X. Hu;H. Hua
Sangyoon Lee;X. Hu;H. Hua
中科院分区:
计算机科学1区
文献类型:
--
作者:
Sangyoon Lee;X. Hu;H. Hua

文献摘要

被引文献

相似文献

在使用光学透视头戴式显示器(OST-HMD)的增强现实环境中,已经探索了关于深度感知问题的许多误差源。然而,两个误差源通常被忽略:光线偏移现象和瞳孔间距(IPD)的变化。第一个误差来源是由OST-HMD所需的光学组合器引起的虚拟和透视光路的折射差异。第二个发生在由于眼睛会聚而引起的观看者的IPD的变化。在本文中,我们分析了这两种误差源对近场深度感知的影响,并提出了补偿这两种误差的方法。此外,我们调查他们的有效性,通过实验比较的条件下,没有我们的误差补偿方法的应用。在我们的实验中,参与者估计自我中心的深度的虚拟和物理对象位于七个不同的近场距离(40~200厘米)使用知觉匹配任务。虽然实验结果显示了不同的模式取决于目标的距离,结果表明,近场深度感知误差可以有效地减少到一个非常小的水平(最多1%的误差),通过补偿上述两个误差源。
Many error sources have been explored in regards to the depth perception problem in augmented reality environments using optical see-through head-mounted displays (OST-HMDs). Nonetheless, two error sources are commonly neglected: the ray-shift phenomenon and the change in interpupillary distance (IPD). The first source of error arises from the difference in refraction for virtual and see-through optical paths caused by an optical combiner, which is required of OST-HMDs. The second occurs from the change in the viewer's IPD due to eye convergence. In this paper, we analyze the effects of these two error sources on near-field depth perception and propose methods to compensate for these two types of errors. Furthermore, we investigate their effectiveness through an experiment comparing the conditions with and without our error compensation methods applied. In our experiment, participants estimated the egocentric depth of a virtual and a physical object located at seven different near-field distances (40~200 cm) using a perceptual matching task. Although the experimental results showed different patterns depending on the target distance, the results demonstrated that the near-field depth perception error can be effectively reduced to a very small level (at most 1 percent error) by compensating for the two mentioned error sources.