The comfort and immersive experience offered by modern near-eye displays is hampered by their technical shortcomings. Such displays produce conflicting visual cues that affect the user's eye movements, creating a condition known as the vergence-accommodation conflict. Vergence is the rotation of the eyes towards each other when looking at a nearby object. Accommodation, on the other hand, refers to a change in the shape of the lens that helps us focus our gaze. In vergence-accommodation conflict, these eye mechanisms do not work together in a natural manner. This results in discomfort, eye fatigue and even nausea for the display user.
There are two approach categories for solving the vergence-accommodation conflict. The first approach is to use an accommodation-invariant display design. In such a display, the link between accommodation and defocus blur is decoupled, so that the viewer sees the image equally sharp regardless of accommodation. This allows the visual signals from vergence to drive accommodation, thus eliminating the discrepancy between them.
The second approach is to use an accommodation-enabling display, where all visual cues are provided accurately and consistently with respect to each other. Such a display creates defocus blur in a natural manner which results in accommodation and vergence operating together without the conflict.
MSc (Tech) Jani Mäkinen demonstrates in his dissertation that both approaches are enhanced by illuminating the displays with coherent light. Coherent light contains temporally stationary light waves with only a single wavelength. Such light can be produced by for example a laser.
The benefits and drawbacks of coherent light
The use of coherent light in near-eye displays can improve their depth of field, thus extending the range of distances where the display viewer sees images at their sharpest. This is useful for the desired response in an accommodation-invariant display. In addition, coherent light enables holography as the means for creating more natural-looking 3D images on an accommodation-enabling display.
The use of coherent light is not without its problems. The perceived images can contain high-contrast speckle noise, which significantly degrades the visual quality. Additionally, holographic imaging methods, or the data generation for 3D imagery, can be computationally heavy.
Current methods in holographic imaging often deliberately degrade the accuracy of the wave models used to generate holograms to reduce the computational burden and improve performance. As a result, the perceived image resolution at far distances from the display is degraded. Alternatively, methods using more accurate wave models can suffer from inaccuracies in modelling of defocus blur from the surfaces of reflective and partially transparent objects. Mäkinen proposes a novel method for computing visually accurate holograms with minimal light modelling approximations. He also examines accelerating the proposed method through machine learning.
The methods developed by Mäkinen use coherent light modelling via wave optics in combination with rigorous signal processing. This approach combination is especially important in reducing speckle noise without introducing additional artefacts, thus improving the observed image quality. Mäkinen also presents a machine learning-based framework to design the optical elements for coherently illuminated accommodation-invariant displays.
“The dissertation results indicate that the use of coherent light in an accommodation-invariant display is an appealing low computational cost solution for solving vergence-accommodation conflict. Additionally, visually accurate holograms can be efficiently generated in real-time using neural networks. These results will be instrumental for the development of the next-generation immersive near-eye displays,” Mäkinen concludes.
Public defence on Thursday 21 December
The doctoral dissertation of MSc (Tech) Jani Mäkinen in the field of signal processing titled Coherent and Holographic Imaging Methods for Immersive Near-Eye Displays will be publicly examined in the Faculty of Information Technology and Communication Sciences at Tampere University in room SA203 in the Sähkötalo building (address: Korkeakoulunkatu 3, Tampere) at 12:00 on Thursday, 21st December 2023
Opponents will be Professor Tomasz Kozacki from Warsaw University of Technology, Poland, and Doctor Ali Özgür Yöntem from University of Cambridge, United Kingdom. The Custos will be Professor Atanas Gotchev from the Faculty of Information Technology and Computing Sciences, Tampere University.
The doctoral dissertation is available online.
The public defence can be followed via remote connection.
Photograph: Robert Bregovic