Seokho Lee
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Seokho Lee has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- Junsuk Rho
Department of Mechanical Engineering, Pohang University of Science and Technology (POSTECH), Pohang 37673, Republic of Korea
5 shared publications - Cherry Park
Department of Mechanical Engineering Pohang University of Science and Technology (POSTECH) Pohang South Korea
2 shared publications - Beomseok Oh
Department of Mechanical Engineering, Pohang University of Science and Technology (POSTECH), Pohang, Republic of Korea
1 shared publication - Chanik Kang
Hanyang University Seoul South Korea
1 shared publication Cherry Park
Department of Mechanical Engineering, Pohang University of Science and Technology (POSTECH), Pohang, Republic of Korea
1 shared publication- Chia-Chun Pan
Department of Electrical and Electronic Engineering, University of Melbourne 2 , Melbourne, VIC
1 shared publication
Latest Publications
- Open Access
- 11 September 2026
- Published by Wiley in Laser & Photonics Reviews
- p. e71877
Compact Single‐Shot Incoherent Edge Detection With a Polarization‐Multiplexed Metalens
Edge detection under incoherent illumination is highly desirable for compact vision systems but remains challenging due to the lack of phase control and the difficulty of implementing high‐pass filtering in intensity‐based imaging. Here, we propose and experimentally demonstrate a polarization‐sensitive metalens that enables single‐shot incoherent edge detection by constructing a second‐order high‐pass optical transfer function (OTF). The metalens imparts distinct phase profiles to left‐ and right‐handed circularly polarized components, producing a pair of spatially separated images that encode complementary object information. Through digital alignment and subtraction, an effective OTF proportional to is realized, leading to robust edge enhancement, while requiring only simple pixel‐wise operations. We experimentally validate this approach using USAF resolution targets and further demonstrate a compact imaging module by directly integrating the metalens with a Complementary Metal‐Oxide‐Semiconductor (CMOS) sensor, enabling real‐scene single‐shot operation. This work establishes a practical optical preprocessing strategy for incoherent edge detection, providing a pathway toward miniaturized, low‐power, and computation‐efficient vision front‐ends for machine vision, autonomous sensing, and portable imaging systems.
- 11 September 2026
- Published by Wiley in Laser & Photonics Reviews
- p. e71877
Open AccessCompact Single‐Shot Incoherent Edge Detection With a Polarization‐Multiplexed Metalens
Edge detection under incoherent illumination is highly desirable for compact vision systems but remains challenging due to the lack of phase control and the difficulty of implementing high‐pass filtering in intensity‐based imaging. Here, we propose and experimentally demonstrate a polarization‐sensitive metalens that enables single‐shot incoherent edge detection by constructing a second‐order high‐pass optical transfer function (OTF). The metalens imparts distinct phase profiles to left‐ and right‐handed circularly polarized components, producing a pair of spatially separated images that encode complementary object information. Through digital alignment and subtraction, an effective OTF proportional to is realized, leading to robust edge enhancement, while requiring only simple pixel‐wise operations. We experimentally validate this approach using USAF resolution targets and further demonstrate a compact imaging module by directly integrating the metalens with a Complementary Metal‐Oxide‐Semiconductor (CMOS) sensor, enabling real‐scene single‐shot operation. This work establishes a practical optical preprocessing strategy for incoherent edge detection, providing a pathway toward miniaturized, low‐power, and computation‐efficient vision front‐ends for machine vision, autonomous sensing, and portable imaging systems.
- 31 July 2026
- Published by Springer Nature in Nature Sensors
- Vol. 1 (8) , 728-739
All-wave computational ultrasonic fingerprint identification with metasurface-driven loop-diffractive neural network
- 31 July 2026
- Published by Springer Nature in Nature Sensors
- Vol. 1 (8) , 728-739
All-wave computational ultrasonic fingerprint identification with metasurface-driven loop-diffractive neural network
- Open Access
- 10 March 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Ultra-efficient physical field computing by complex-valued network quantization
Neural network quantization is an established technique for compressing real-valued models, but its application to complex-valued networks—essential in electromagnetics, acoustics, and quantum physics—remains underdeveloped. Conventional quantization methods treat real and imaginary components as independent channels, thereby disrupting the algebraic structure of complex multiplication and distorting essential phase relationships. To address this problem, we propose a real-imaginary joint quantization method that minimizes error propagation in complex multiplication and maintains coherence in phase-sensitive tasks, thereby preserving amplitude-phase fidelity during complex-valued inference. Combined with physics-aware adaptive precision training, this approach demonstrates outstanding performance across hologram generation, audio classification, wireless signal classification, and synthetic aperture radar signal recognition tasks. Compared to the state-of-the-art hologram generation model HoloNet, our approach achieves a 3.9 dB improvement in peak signal-to-noise ratio while reducing computational load and memory consumption by 99.1% and 99.8%, respectively. This research establishes a pathway toward lightweight, high-fidelity complex-valued neural networks for scientific computing and coherent signal processing.
- 10 March 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Open AccessUltra-efficient physical field computing by complex-valued network quantization
Neural network quantization is an established technique for compressing real-valued models, but its application to complex-valued networks—essential in electromagnetics, acoustics, and quantum physics—remains underdeveloped. Conventional quantization methods treat real and imaginary components as independent channels, thereby disrupting the algebraic structure of complex multiplication and distorting essential phase relationships. To address this problem, we propose a real-imaginary joint quantization method that minimizes error propagation in complex multiplication and maintains coherence in phase-sensitive tasks, thereby preserving amplitude-phase fidelity during complex-valued inference. Combined with physics-aware adaptive precision training, this approach demonstrates outstanding performance across hologram generation, audio classification, wireless signal classification, and synthetic aperture radar signal recognition tasks. Compared to the state-of-the-art hologram generation model HoloNet, our approach achieves a 3.9 dB improvement in peak signal-to-noise ratio while reducing computational load and memory consumption by 99.1% and 99.8%, respectively. This research establishes a pathway toward lightweight, high-fidelity complex-valued neural networks for scientific computing and coherent signal processing.
- Open Access
- 2 March 2026
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 139 (9)
3D-printed mechanically reconfigurable all-dielectric metagrating for terahertz filters
We demonstrate a mechanically reconfigurable terahertz filter implemented by the 3D-printed all-dielectric metagrating, providing a versatile platform for adaptive photonic devices. The structure consists of parallel supporting bars connected by vertical grating elements, forming a lattice that can be mechanically reconfigured. By applying a controlled lateral displacement to the supporting bars, the grating elements are tilted, effectively reducing the grating period and inducing a systematic blue shift in the transmission dip frequency. By tilting the metagrating, we experimentally observed a 15% frequency modulation and a 95% transmission modulation. The numerical simulations show consistent results with the experimental results. In contrast to conventional approaches that rely on refractive index modulation, phase-change materials, or other active tuning mechanisms, this method achieves tunability purely through geometric reconfiguration. This strategy offers a simple, low-cost, and robust route to dynamic terahertz filtering, and it can be readily extended to other reconfigurable metamaterials and photonic devices requiring adaptive control of their optical response.
- 2 March 2026
- Published by AIP Publishing in Journal of Applied Physics
- Vol. 139 (9)
Open Access3D-printed mechanically reconfigurable all-dielectric metagrating for terahertz filters
We demonstrate a mechanically reconfigurable terahertz filter implemented by the 3D-printed all-dielectric metagrating, providing a versatile platform for adaptive photonic devices. The structure consists of parallel supporting bars connected by vertical grating elements, forming a lattice that can be mechanically reconfigured. By applying a controlled lateral displacement to the supporting bars, the grating elements are tilted, effectively reducing the grating period and inducing a systematic blue shift in the transmission dip frequency. By tilting the metagrating, we experimentally observed a 15% frequency modulation and a 95% transmission modulation. The numerical simulations show consistent results with the experimental results. In contrast to conventional approaches that rely on refractive index modulation, phase-change materials, or other active tuning mechanisms, this method achieves tunability purely through geometric reconfiguration. This strategy offers a simple, low-cost, and robust route to dynamic terahertz filtering, and it can be readily extended to other reconfigurable metamaterials and photonic devices requiring adaptive control of their optical response.
- Open Access
- 1 October 2025
- Published by AIP Publishing in APL Photonics
- Vol. 10 (10)
Inverse design for enhanced chiroptical response with chiral nanophotonic structures
Chiral nanophotonic structures have attracted significant attention due to their wide-ranging applications in chiral sensing, on-chip optical devices, and optical communication. Although most chiral structures to date have been generated using conventional design approaches—guided by human intuition and followed by parameter sweeps—these methods are inherently limited in efficiently exploring complex, high-dimensional design spaces. Recently, inverse design methods have emerged to overcome traditional limitations by systematically and efficiently navigating vast parameter spaces to discover non-intuitive chiral geometries with enhanced performance. In this work, we present a range of chiral nanophotonic structures along with their corresponding design approaches, as well as performance metrics. This perspective focuses on inverse design methodologies, such as metaheuristic optimization, topology optimization, and data-driven approaches, tailored to the needs of enhanced chiroptical response. We also introduce conventional design strategies and compare them with optimization- and learning-based design pipelines. Particular emphasis is placed on the formulation of objective functions for circular dichroism and helical dichroism sensing, structural symmetry breaking, and computational constraints in structure design. We further highlight representative high-performance chiral metasurface and metamaterial designs developed using these strategies and conclude by outlining open challenges and future research directions.
- 1 October 2025
- Published by AIP Publishing in APL Photonics
- Vol. 10 (10)
Open AccessInverse design for enhanced chiroptical response with chiral nanophotonic structures
Chiral nanophotonic structures have attracted significant attention due to their wide-ranging applications in chiral sensing, on-chip optical devices, and optical communication. Although most chiral structures to date have been generated using conventional design approaches—guided by human intuition and followed by parameter sweeps—these methods are inherently limited in efficiently exploring complex, high-dimensional design spaces. Recently, inverse design methods have emerged to overcome traditional limitations by systematically and efficiently navigating vast parameter spaces to discover non-intuitive chiral geometries with enhanced performance. In this work, we present a range of chiral nanophotonic structures along with their corresponding design approaches, as well as performance metrics. This perspective focuses on inverse design methodologies, such as metaheuristic optimization, topology optimization, and data-driven approaches, tailored to the needs of enhanced chiroptical response. We also introduce conventional design strategies and compare them with optimization- and learning-based design pipelines. Particular emphasis is placed on the formulation of objective functions for circular dichroism and helical dichroism sensing, structural symmetry breaking, and computational constraints in structure design. We further highlight representative high-performance chiral metasurface and metamaterial designs developed using these strategies and conclude by outlining open challenges and future research directions.