Mass Eye and Ear Researchers Suggest Ways to Improve Visual Perception in Virtual Reality

Smart glasses and head-mounted displays (HMDs) for virtual reality make use of vision multiplexing, an optical engineering approach that superimposes augmented information over an observer's natural field of view. First-generation HMDs incorporate an opaque panel mounted over just one eye ("monocular display"), and part of the background scene is occluded.

Subscribe to the latest updates from Ophthalmology Advances in Motion

Error: Please enter a valid email address.

Email Address Submit

Newer HMDs have "see-through displays," which present the augmented information and the transmitted background scene simultaneously. The Google Glass and Eye Tap device provide monocular augmented information, and binocular see-through HMDs were recently developed to present stereoscopic augmented information on top of the background scene. Examples are the Microsoft HoloLens and Epson Moverio smart glasses and the Magic Leap HMD.

The capability to provide multiple sources of information within the same field of view could allow vision multiplexing to be useful during movement, such as walking, driving and landing aircraft in adverse weather conditions with low visibility. The technology can also expand the field of view for patients with visual field loss.

However, vision multiplexing can lead to unintended perceptual experiences, including visual rivalry—visual perception alternates between the superimposed augmented image and the background scene. This may interfere with the application of vision multiplexing to tasks requiring movement.

Researchers at Mass Eye and Ear and the Agency for Science, Technology and Research in Singapore investigated how different vision multiplexing configurations affect detection of augmented information during movement. Jae-Hyun Jung, PhD, assistant scientist at Schepens Eye Research Institute of Mass Eye and Ear, Sujin Kim, PhD, a research fellow at Schepens Eye Research Institute, and Shui'er Han, PhD, a scientist at the Agency for Science, Technology and Research, present the results and discuss recommendations for device and application design in Scientific Reports.

Methods

19 individuals, ages 23 to 51, were recruited for the study. All had normal or corrected-to-normal visual acuity and normal stereo vision.

The effects of three factors—eccentricity of the peripheral displays, depth condition and eye movement—were examined in separate experiments. All experiments tested monocular opaque, mocnoular see-through, and binocular see-through multiplexing configurations in the periphery and simulated the optic flow of forward walking in a 3D virtual corridor. The researchers measured the visibility of the augmented target and not the competing image.

Experiment 1: Eccentricity (n=12)

Both eccentricity and multiplexing configuration modulated target visibility. The binocular see-through target was significantly more visible than the monocular opaque and see-through targets, suggesting it is most resistant to binocular rivalry.

Furthermore, the monocular target was significantly more visible when it was opaque than when it was see-through. This suggests suppression of the target by monocular rivalry between the augmented foreground and the see-through moving background.

The visibility of the binocular see-through target was higher than that of the monocular opaque target despite being of lower contrast. The contribution of binocular rivalry to target visibility seems to be more important than target contrast.

Experiment 2: Depth Condition (n=13)

As in experiment 1, the binocular see-through target was more visible than the monocular opaque and monocular see-through targets. Also similar to experiment 1, target visibility was lower in the monocular see-through configuration than in the monocular opaque configuration, presumably due to monocular rivalry and lower target contrast.

The target became significantly less visible when the binocular see-through configuration was presented with the 3D depth condition than the 2D condition. The monocular see-through configuration was also less visible in the 3D condition, but the difference was not statistically significant.

Experiment 3: Eye Movements (n=13)

Similar to the results of experiments 1 and 2, the binocular see-through target was most visible and the monocular opaque target was more visible than the monocular see-through target regardless of eye movements.

In all three multiplexing configurations, the target was more visible when saccadic or smooth pursuit eye movements were being executed than when the eye was fixed at one location.

Future Directions

The binocular see-through display appears to be most resistant to visual rivalry suppression but remains susceptible to depth switching, although that would probably be reduced in real-world viewing conditions.

The optimal approach may be to minimize rivalry effects in monocular displays, which should be achievable by manipulating the display contrast. The contrast manipulation might have to be dynamic—further analyses showed that increases in display contrast (monocular opaque vs. monocular see-through) increased overall visibility but not the average duration of each period when the target was visible.

Some possible options are to introduce contrast transients in the display to maintain its dominance over the conflicting background area or to modulate the display contrast, keeping in mind how vestibular signals affect visual perception.

view original journal article Subscription may be required

Learn more about research in Ophthalmology at Mass Eye and Ear/Mass General

Refer a patient to the Mass Eye and Ear/Mass General