Effects of Configuration of Optical Combiner on Near-Field Depth Perception in Optical See-Through Head-Mounted Displays

• Published in: IEEE transactions on visualization and computer graphics Vol. 22; no. 4; pp. 1432 - 1441
• Main Authors: Lee, Sangyoon, Hua, Hong
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.04.2016; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptive optics; augmented reality; Configurations; Helmet mounted displays; near-field depth perception; optical combiner; Optical distortion; Optical imaging; Optical paths; Optical refraction; optical see-through head-mounted display; Optical variables control; Optical waveguides; Ray-shift phenomenon; Space perception; Three-dimensional displays; Waveguides; augmented reality; Ray-shift phenomenon; near-field depth perception; optical combiner; optical see-through head-mounted display
• ISSN: 1077-2626; 1941-0506
• DOI: 10.1109/TVCG.2016.2518138

The ray-shift phenomenon means the apparent distance shift in the display image plane between virtual and physical objects. It is caused by the difference in the refraction of virtual display and see-through optical paths derived from optical combiners that are necessary to provide a see-through capability in optical see-through head-mounted displays. In this work, through a human-subject experiment, we investigated the effects of ray-shift phenomenon induced by the optical combiner on depth perception for near-field distances (40 cm-100 cm). In our experiment, we considered three different configurations of optical combiner: horizontal-tilt and vertical-tilt configurations (using plate beamsplitters horizontally and vertically tilted by 45°, respectively), and non-tilt configuration (using rectangular solid waveguides). Participants' depth perception errors in these configurations were compared with those in an ordinary condition (i.e., the condition where physical objects are directly shown without the displays) and theoretically estimated ones. According to the experimental results, the measured percentage depth perception errors were similar to the theoretically estimated ones, where the amount of estimated percentage depth errors was greater than 0.3%. Furthermore, the participants showed significantly larger depth perception errors in the horizontal-tilt configuration than in an ordinary condition, while no large errors were found in the vertical-tilt configuration. In the non-tilt configuration, the results were dependent on the thickness of optical combiner and target distance.