Physics-guided deep learning-based real-time image reconstruction of Fourier-domain optical coherence tomography DOI Creative Commons
Mengyuan Wang, Jianing Mao, Hang Su

и другие.

Biomedical Optics Express, Год журнала: 2024, Номер 15(11), С. 6619 - 6619

Опубликована: Окт. 24, 2024

In this paper, we introduce a physics-guided deep learning approach for high-quality, real-time Fourier-domain optical coherence tomography (FD-OCT) image reconstruction. Unlike traditional supervised methods, the proposed method employs unsupervised learning. It leverages underlying OCT imaging physics to guide neural networks, which could thus generate high-quality images and provide physically sound solution original problem. Evaluations on synthetic experimental datasets demonstrate superior performance of our approach. The achieves highest quality metrics compared inverse discrete Fourier transform (IDFT), optimization-based several state-of-the-art methods based Our enables frame rates 232 fps 87 images, represents significant improvements over existing techniques. learning-based offer promising FD-OCT reconstruction, potentially paves way leveraging power in real-world applications.

Язык: Английский

End-to-end learning of 3D phase-only holograms for holographic display DOI Creative Commons
Liang Shi, Beichen Li, Wojciech Matusik

и другие.

Light Science & Applications, Год журнала: 2022, Номер 11(1)

Опубликована: Авг. 3, 2022

Computer-generated holography (CGH) provides volumetric control of coherent wavefront and is fundamental to applications such as 3D displays, lithography, neural photostimulation, optical/acoustic trapping. Recently, deep learning-based methods emerged promising computational paradigms for CGH synthesis that overcome the quality-runtime tradeoff in conventional simulation/optimization-based methods. Yet, quality predicted hologram intrinsically bounded by dataset's quality. Here we introduce a new dataset, MIT-CGH-4K-V2, uses layered depth image data-efficient input two-stage supervised+unsupervised training protocol direct high-quality phase-only holograms. The proposed system also corrects vision aberration, allowing customization end-users. We experimentally show photorealistic holographic projections discuss relevant spatial light modulator calibration procedures. Our method runs real-time on consumer GPU 5 FPS an iPhone 13 Pro, drastically enhanced performance above.

Язык: Английский

Процитировано

102
4K-DMDNet: diffraction model-driven network for 4K computer-generated holography DOI Creative Commons
Ke‐Xuan Liu, Jiachen Wu, Zehao He

и другие.

Opto-Electronic Advances, Год журнала: 2023, Номер 6(5), С. 220135 - 220135

Опубликована: Янв. 1, 2023

Deep learning offers a novel opportunity to achieve both high-quality and high-speed computer-generated holography (CGH). Current data-driven deep algorithms face the challenge that labeled training datasets limit performance generalization. The model-driven introduces diffraction model into neural network. It eliminates need for dataset has been extensively applied hologram generation. However, existing problem of insufficient constraints. In this study, we propose network capable high-fidelity 4K generation, called Diffraction Model-driven Network (4K-DMDNet). constraint reconstructed images in frequency domain is strengthened. And structure combines residual method sub-pixel convolution built, which effectively enhances fitting ability inverse problems. generalization 4K-DMDNet demonstrated with binary, grayscale 3D images. High-quality full-color optical reconstructions holograms have achieved at wavelengths 450 nm, 520 638 nm.

Язык: Английский

Процитировано

75
Holographic tomographic volumetric additive manufacturing DOI Creative Commons
Maria Isabel Álvarez-Castaño, Andreas Erik Gejl Madsen, Jorge Madrid‐Wolff

и другие.

Nature Communications, Год журнала: 2025, Номер 16(1)

Опубликована: Фев. 11, 2025

Язык: Английский

Процитировано

3
Deep-Learning Computational Holography: A Review DOI Creative Commons
Tomoyoshi Shimobaba, David Blinder, Tobias Birnbaum

и другие.

Frontiers in Photonics, Год журнала: 2022, Номер 3

Опубликована: Март 28, 2022

Deep learning has been developing rapidly, and many holographic applications have investigated using deep learning. They shown that can outperform previous physically-based calculations lightwave simulation signal processing. This review focuses on computational holography, including computer-generated holograms, displays, digital We also discuss our personal views the promise, limitations future potential of in holography.

Язык: Английский

Процитировано

59
Non-convex optimization for inverse problem solving in computer-generated holography DOI Creative Commons
Xiaomeng Sui, Zehao He, Daping Chu

и другие.

Light Science & Applications, Год журнала: 2024, Номер 13(1)

Опубликована: Июль 9, 2024

Abstract Computer-generated holography is a promising technique that modulates user-defined wavefronts with digital holograms. Computing appropriate holograms faithful reconstructions not only problem closely related to the fundamental basis of but also long-standing challenge for researchers in general fields optics. Finding exact solution desired hologram reconstruct an accurate target object constitutes ill-posed inverse problem. The practice single-diffraction computation synthesizing can provide approximate answer, which subject limitations numerical implementation. Various non-convex optimization algorithms are thus designed seek optimal by introducing different constraints, frameworks, and initializations. Herein, we overview applied computer-generated holography, incorporating principles synthesis based on alternative projections gradient descent methods. This aimed underlying optimized generation, as well insights into cutting-edge developments this rapidly evolving field potential applications virtual reality, augmented head-up display, data encryption, laser fabrication, metasurface design.

Язык: Английский

Процитировано

17
Ergonomic‐Centric Holography: Optimizing Realism, Immersion, and Comfort for Holographic Display DOI Creative Commons
Liang Shi, DongHun Ryu, Wojciech Matusik

и другие.

Laser & Photonics Review, Год журнала: 2024, Номер 18(4)

Опубликована: Фев. 25, 2024

Abstract Ergonomic‐centric holography is introduced, an algorithmic framework that simultaneously optimizes for realistic incoherent defocus, unrestricted pupil movements in the eye box, and high‐order diffractions filtering‐free holography. The proposed method outperforms prior algorithms on holographic display prototypes operating unfiltered pupil‐mimicking modes, offering potential to enhance next‐generation virtual augmented reality experiences.

Язык: Английский

Процитировано

11
Holographic Glasses for Virtual Reality DOI
Jonghyun Kim, Manu Gopakumar, Suyeon Choi

и другие.

Опубликована: Июль 20, 2022

We present Holographic Glasses, a holographic near-eye display system with an eyeglasses-like form factor for virtual reality. Glasses are composed of pupil-replicating waveguide, spatial light modulator, and geometric phase lens to create images in lightweight thin factor. The proposed design can deliver full-color 3D using optical stack 2.5 mm thickness. A novel pupil-high-order gradient descent algorithm is presented the correct calculation user's varying pupil size. implement benchtop wearable prototypes testing. Our binocular prototype supports focus cues provides diagonal field view 22.8° 2.3 static eye box additional capabilities dynamic beam steering, while weighing only 60 g excluding driving board.

Язык: Английский

Процитировано

36
3D-HoloNet: Fast, unfiltered, 3D hologram generation with camera-calibrated network learning DOI
Wenbin Zhou, Feifan Qu, Xiangyue Meng

и другие.

Optics Letters, Год журнала: 2025, Номер 50(4), С. 1188 - 1188

Опубликована: Янв. 8, 2025

Computational holographic displays typically rely on time-consuming iterative computer-generated (CGH) algorithms and bulky physical filters to attain high-quality reconstruction images. This trade-off between inference speed image quality becomes more pronounced when aiming realize 3D imagery. work presents 3D-HoloNet , a deep neural network-empowered CGH algorithm for generating phase-only holograms (POHs) of scenes, represented as RGB-D images, in real time. The proposed scheme incorporates learned, camera-calibrated wave propagation model phase regularization prior into its optimization. unique combination allows accommodating practical, unfiltered display setups that may be corrupted by various hardware imperfections. Results tested an reveal the can achieve 30 fps at full HD one color channel using consumer-level GPU while maintaining comparable methods across multiple focused distances.

Язык: Английский

Процитировано

1
Investigating learning-empowered hologram generation for holographic displays with ill-tuned hardware DOI
Xinxing Xia,

Furong Yang,

Weisen Wang

и другие.

Optics Letters, Год журнала: 2023, Номер 48(6), С. 1478 - 1478

Опубликована: Фев. 13, 2023

Existing computational holographic displays often suffer from limited reconstruction image quality mainly due to ill-conditioned optics hardware and hologram generation software. In this Letter, we develop an end-to-end hardware-in-the-loop approach toward high-quality for displays. Unlike other methods using ideal wave propagation, ours can reduce artifacts introduced by both the light propagation model setup, in particular non-uniform illumination. Experimental results reveal that, compared with classical computer-generated algorithm counterparts, better of images be delivered without a strict requirement on fine assembly optical components good uniformity laser sources.

Язык: Английский

Процитировано

11
High-quality Phase-only Fourier Hologram Generation with Camera-in-the-loop DOI Creative Commons
Han‐Ju Yeom, Keehoon Hong, Minsik Park

и другие.

Optics Express, Год журнала: 2025, Номер 33(4), С. 6615 - 6615

Опубликована: Янв. 31, 2025

In this paper, we propose an optimization method for Fourier holograms that enables high-quality optical reconstruction of phase-only holograms. We define the amplitude input image hologram calculation as plane within a camera-in-the-loop (CITL) framework to generate with superior quality. Unlike traditional CITL methods optimize phase holograms, our proposed optimizes amplitudes exhibit high correlation original images in hologram. Leveraging correlation, introduce neural network model hologram, PoFNet, infer optimized from images, thereby addressing time-consuming nature algorithm, which is hindered by repetitive calculations. During training process, PoFNet employs account non-ideal forward propagation, i.e., propagation. Optical experiments demonstrate both and effectively reduce noise path.

Язык: Английский

Процитировано

0