Improved vergence and accommodation via Purkinje Image tracking with multiple cameras for AR glasses

Lu, Conny; Chakravarthula, Praneeth; Tao, Yujie; Chen, Steven; Fuchs, Henry

doi:10.1109/ismar50242.2020.00058

Cited by 19 publications

(12 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Vergence can be estimated by using two-dimensional pupil size changes only, but such methods are sensitive to ambient illumination changes and require calibration for each user 29,30 . In our work, we investigated the effect of pupil parameters on the vergence and accommodation predictions.…”

Section: Accommodation and Vergence Prediction Utilizing Machine Lear...mentioning

confidence: 99%

“…Eye-trackers employ P4 since it is more easy to detect due to its brightness and location on the camera. Recent studies employed P3 and have stated that it outperforms state-of-the-art eye trackers, utilizing deep learning models for their estimation 29,30 . To quantify the effects of the Purkinje image locations on the accommodation and vergence estimations, we used a feature importance technique on the whole dataset (experimental results from 9 subjects) named permutation importance.…”

Section: Depth Error [D] Angle Errormentioning

confidence: 99%

“…With the rise of the machine learning and deep learning techniques, researchers have explored the use of these methods on Purkinje reflections to develop an eye tracker that can also predict accommodation depth [29][30][31] . However, these studies use either P1 with P3 29,30 or P1 with P4 27,28,31 , while discarding the other reflection. As a result, data and the accuracy of model predictions have been limited.In our previous study, we proposed a primitive version of our work, illustrating that accommodation can be measured dynamically using Purkinje images for different subjects 32 .…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Dynamic accommodation measurement using Purkinje reflections and machine learning

Ozhan

Aygun

Şahin

et al. 2023

Preprint

View full text Add to dashboard Cite

Quantifying eye movement is important for diagnosing various neurological and ocular diseases as well as AR/VR displays. We developed a simple setup for real-time dynamic gaze tracking and accommodation measurements based on Purkinje reflections, which are the reflections from front and back surfaces of the cornea and the eye lens. We used an accurate eye model in ZEMAX to simulate the Purkinje reflection positions at different focus distances of the eye, which matched the experimental data. A neural network was trained to simultaneously predict vergence and accommodation using data collected from 9 subjects as well as synthetic data generated using the eye model. We demonstrated that the use of Purkinje reflection coordinates together with pupil center and shape in machine learning improves the accuracy of the prediction. The proposed system accurately predicted the accommodation with an accuracy better than 0.21D using two-point calibration data in tests performed with 9 subjects in our setup.

show abstract

Section: Accommodation and Vergence Prediction Utilizing Machine Lear...mentioning

confidence: 99%

Section: Depth Error [D] Angle Errormentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Dynamic accommodation measurement using Purkinje reflections and machine learning

Ozhan

Aygun

Şahin

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The VAC causes viewer fatigue and discomfort as well as distortions in the perceived depth [24,72]. Much research has been carried out in an attempt to solve the VAC issue in both VR and AR displays, which include varifocal displays (VR [32,36,38,55] and AR [1,10,15,18,30,75]), multifocal displays (VR [3,13,14,25,43,[50][51][52]61] and AR [4,16,41,42,44,60,65] ), light field displays (VR [23,26,39] and AR [46,48]), holographic displays (VR [11,12,49,58] and AR [30,47] ), and other display techniques, such as those that use color information [2,17]. Unfortunately, despite the substantial effort that has been put into solving the VAC for stereoscopic flat-panel displays and HMDs, researchers have not focused on the issue for stereoscopic PM.…”

Section: Mitigating the Vacmentioning

confidence: 99%

Multifocal Stereoscopic Projection Mapping

Kimura,

Iwai,

Punpongsanon

et al. 2021

Preprint

View full text Add to dashboard Cite

Fig. 1. The proposed multifocal stereoscopic projection mapping technique mitigates the vergence-accommodation conflict by placing the virtual image (as an optical term) at the desired distance from an observer using focal-sweep active shutter glasses and a synchronized high-speed projector. (a) Prototype system; (b) experimental setup to show the multifocal appearance of Stanford bunnies onto a flat surface (D) whose virtual images were placed, from near to far, at points at which three physical objects (A-C) were placed; (c-e) captured scene observed through a focal sweep lens. The bunnies, as well as the physical objects, appeared focused from the bottom to the top according to the focus distance of the camera lens at (c) near, (d) middle, and (e) far.

show abstract

“…Lu et al. proposed a high-precision eye tracker based on multi-camera and multi-LED to improve AR glasses’ adjustment and gaze estimation by effectively tracking sensitive higher-order Purkinje images 7 . This study showed that the relative sensitivity of Purkinje image III with the change of fixation relative to the center of the pupil was the highest.…”

Section: Introductionmentioning

confidence: 98%

Gaze distance detection by binocular tracking for automatic zoom glasses

et al. 2023

View full text Add to dashboard Cite

.To realize automatic zooming of presbyopia glasses, we proposed a gaze distance detection method based on binocular tracking. A theoretical model was built to calculate gaze distance based on the gaze directions of two eyeballs. The gaze direction of each eyeball was obtained by the pupil center corneal reflection method. Three light-emitting diodes (LEDs) and two miniature cameras were placed on the glasses frame around the eyeball. Then, gaze direction was derived by calculating the distances between the pupil center and the positions of three LED reflection images by the corneal from the images captured by cameras. The experiment setup was built to test the gaze direction and gaze distance. The experimental result demonstrates that it is an effective way to obtain the user’s gaze distance for automatically adjusting the focal length of the zoom glasses.

show abstract

Improved vergence and accommodation via Purkinje Image tracking with multiple cameras for AR glasses

Cited by 19 publications

References 52 publications

Dynamic accommodation measurement using Purkinje reflections and machine learning

Dynamic accommodation measurement using Purkinje reflections and machine learning

Multifocal Stereoscopic Projection Mapping

Gaze distance detection by binocular tracking for automatic zoom glasses

Contact Info

Product

Resources

About