▲ Professor Jae-Hyeung Park (corresponding author), Ph.D. candidate Woongseob Han (first author), and integrated M.S.–Ph.D. candidate Chanseul Lee (co-author), Department of Electrical and Computer Engineering, Seoul National University

The College of Engineering at Seoul National University announced that a research team led by Professor Jae-Hyeung Park from the ECE Department has developed an occlusion-based holographic augmented reality (AR) display that significantly enhances visual realism in AR environments.

By integrating a holographic display with an optical occlusion system, the team achieved a new level of visual fidelity in AR. Furthermore, they implemented opaque three-dimensional virtual imagery and optically generated virtual shadows, reproducing visual effects in which virtual objects interact with real-world environments.

Recognizing that visual information in AR environments is typically concentrated around virtual objects rather than distributed across the entire space, the research team also introduced an AI-based hologram generation algorithm optimized for sparse holographic images*.

* sparse holographic image: a hologram in which visual data exists only in limited regions of the entire image space

On October 7, the findings were published as an Inside Front Cover Article in Laser & Photonics Reviews (IF: 10.0), an internationally renowned journal in the field of optics published by Wiley-VCH based in Germany.

▲ Laser & Photonics Reviews Inside Front Cover

■ Research Background

As AR glasses emerge as next-generation smart devices following smartphones, leading global technology companies have been actively investing in their development. Despite rapid progress, however, currently commercialized or publicly unveiled AR glasses remain limited in that they are unable to implement occlusion effects, where virtual images obscure real-world objects. The absence of occlusion—a key visual cue for human depth perception—causes virtual images to appear semi-transparent and unrealistically superimposed on the physical environment. Consequently, the realism of the AR environment and the user’s sense of immersion are significantly diminished.

Moreover, existing AR glasses reproduce three-dimensional images solely through binocular disparity while keeping monocular depth cues fixed, resulting in a vergence-accommodation conflict (VAC). The visual fatigue and dizziness experienced by users due to VAC have long been identified as major obstacles to the widespread adoption of near-eye displays, including AR glasses.

Previous studies addressing this issue have explored approaches such as adding an occlusion optics module in front of the display to selectively block real-world light and render opaque virtual images, or employing techniques such as holography, light-field displays, and variable-focus technologies to reproduce monocular three-dimensional images. However, efforts to achieve both occlusion effects and three-dimensional imaging simultaneously have largely remained at a preliminary stage, underscoring the need for more in-depth research to enhance the visual realism of AR environments.

■ Research Achievements

To address this challenge, Prof. Jae-Hyeung Park’s research team developed a holographic AR display that realizes an AR environment with unprecedented visual realism by combining a holographic AR system—capable of reproducing ideal three-dimensional images—with an optical occlusion system that blocks the real-world background.

The research team first noted that the structure of the optical occlusion system is identical to that of a 4f system-based Fourier filter structure*, which is commonly used in holographic displays to eliminate noise information. Accordingly, the team utilized a single Digital Micromirror Device (DMD)* placed within a single 4f system to function as both a Fourier filter and an occlusion mask. By employing a time-multiplexing technique, they enabled both occlusion and noise elimination to be performed within a unified system.

* Fourier filter structure : optical configuration that analyzes complex images into their frequency components, allowing selective removal or correction of noise

* DMD : reflective optical element composed of numerous microscopic mirrors, used to rapidly control the brightness and pattern of holographic display images

Taking a step further, the research team incorporated the dynamic characteristics of the DMD—unlike conventional fixed Fourier filters—into the AI-based hologram generation algorithm. This approach significantly reduced the search space of the optimizer, achieving an average improvement of 11 decibels (dB) in the Peak Signal-to-Noise Ratio (PSNR)* of sparse holographic images compared to conventional algorithms under identical conditions. In addition, by applying a time-multiplexing technique, the team suppressed Speckle noise*—a major factor degrading holographic image quality—and doubled the field of view.

* PSNR : a measure of the ratio between the original and reconstructed image signals, used to evaluate image or image reconstruction quality. A higher PSNR value indicates superior image quality.

* Speckle noise : a granular interference pattern that appears as fine dotted noise in images

The research team also built a benchtop prototype* based on the proposed system, successfully reproducing opaque three-dimensional AR images in which virtual objects occlude the real-world background. Further, the team reproduced AR scenes in which virtual objects cast realistic shadows onto the real world by leveraging the occlusion effect. Experimental results showed a significant increase in contrast and image sharpness compared to conventional AR displays without occlusion, marking the world’s first realization of high-contrast, high-fidelity three-dimensional AR scenes free from background interference.

* benchtop prototype : a small-scale experimental device built prior to commercialization to verify a system’s performance and operating principles

▲ Experimental results showing a holographic 3D image that occludes the real background and casts a shadow

■ Expected Impact

This research is significant in that it realized a true form of AR in which virtual images optically interact with the real environment. The proposed technology enables virtual objects to selectively block real-world light and cast shadows, establishing itself as a next-generation display technology capable of delivering an AR experience that feels natural to human visual perception.

Additionally, this achievement marks a departure from conventional hologram optimization methods that have primarily relied on software-based methods. By directly integrating a dynamically operating Fourier filter into the algorithmic framework, the research introduced a new paradigm in which physical hardware enhances algorithmic performance. Highlighting the potential of hardware-algorithm co-design, this technology is expected to find applications in next-generation immersive display systems.

■ Researchers' Remarks

Prof. Jae-Hyeung Park, who supervised the research, stated, “This study demonstrates the potential for a new form of AR in which virtual images interact with light of the real environment,” adding, “We will continue to develop next-generation display technologies that provide more natural and immersive visual experiences through the convergence of optics and artificial intelligence.”

■ Researcher Career Path

First author Woongseob Han is currently pursuing his Ph.D. in the ECE Department at SNU, where he continues his research on near-eye displays for AR/VR and next-generation three-dimensional display technologies. After graduation, he plans to work as an optical design engineer in the field of next-generation immersive displays at research institutes or companies in Korea and abroad.

[Reference]

- Paper Title/Journal : “Enhancing Realism in Holographic Augmented Reality Displays Through Occlusion Handling”, Laser & Photonics Reviews
- DOI : https://doi.org/10.1002/lpor.202501052

[Contact]

Prof. Jae-Hyeung Park, Three-Dimensional Optical Engineering Laboratory, Department of Electrical and Computer Engineering, Seoul National University / 02-880-1825 / jaehyeung@snu.ac.kr

Source: https://ece.snu.ac.kr/ece/news?md=v&bbsidx=57000

Translated by: Changhoon Kang, English Editor of the Department of Electrical and Computer Engineering, changhoon27@snu.ac.kr