Lensless Magnified Holographic Projection Based on an Unsupervised Neural Network Technology

Xingjian Li,Ping Su,Bizhong Xia

DOI: https://doi.org/10.1364/dh.2022.w5a.29

2022-01-01

Abstract:A new method is proposed to generate amplitude-only holograms in a lensless magnified projection system based on the unsupervised neural network technology. Simulation demonstrated that the method can generate high-quality holograms in 0.22s.

What problem does this paper attempt to address?

All-optical Diffractive Neural Networked Terahertz Hologram.

Dashuang Liao,Ka Fai Chan,Chi Hou Chan,Qingle Zhang,Haogang Wang

DOI: https://doi.org/10.1364/ol.394046

2020-01-01

Abstract:Holography has garnered an explosion of interest in tremendous applications, owing to its capability of storing amplitude and phase of light and reconstructing the full-wave information of targets. Spatial light modulators, metalenses, metasurfaces, and other devices have been explored to achieve holographic images. However, the required phase distributions for conventional holograms are generally calculated using the Gerchberg-Saxton algorithm, and the iteration is time-consuming without Fourier transform or other acceleration techniques. Few studies on designing holograms using artificial intelligence methods have been conducted. In this Letter, a three-dimensional (3D)-printed hologram for terahertz (THz) imaging based on a diffractive neural network (DNN) is proposed. Target imaging letters "THZ" with uniform field amplitude are assigned to a predefined imaging surface. Quantified phase profiles are primarily obtained by training the DNN with the target image and input field pattern. The entire training process takes only 60 s. Consequently, the hologram, that is, a two-dimensional array of dielectric posts with variational heights that store phase information, is fabricated using a 3D printer. The full-wave simulation and experimental results demonstrate the capability of the proposed hologram to achieve high-quality imaging in the THz regime. The proposed lens and design strategy may open new possibilities in display, optical-data storage, and optical encryption.
Lensless holographic dynamic projection system based on weakly supervised learning

Yaping Huang,Junrong Wang,Ping Su,Jianshe Ma

DOI: https://doi.org/10.1016/j.optlastec.2024.111219

IF: 4.939

2024-05-30

Optics & Laser Technology

Abstract:Diffraction-based lensless projection system offers the advantages of a compact size with high flexibility. However, existing iterative methods of computation for computer-generated holograms (CGHs) in projection systems are not mature enough. Furthermore, slow CGH generation process is not suitable in dynamic display pattern. Here, a lensless projection system with large field of white light without chromatic aberration is proposed, enabling rapid hologram generation and white light display utilizing a single hologram under trichromatic illumination. The core innovation lies in leveraging differences in reconstructed images from holograms acquired under different wavelength conditions to constrain network training. Additionally, a lightweight hologram generation network is designed for hologram synthesis. This network is trained in a weakly supervised manner, utilizing the Gerchber-Saxton (GS) iterative algorithm based on the scaled angular spectrum method (SASM–GS) to generate reference holograms. The utilization of a weakly supervised approach enhances the efficiency of network training and overcomes the limitations imposed by label-based training. Simulation results demonstrate that the proposed method can rapidly generate holograms with low chromatic aberration, achieving a hologram generation speed of 40 frames per second while maintaining a peak signal-to-noise ratio of 35 dB for reconstructed images. A lensless projection system based on digital micromirror device (DMD) is constructed, and optical experiments confirm that the proposed method enables color-consistent white light display using a single hologram, making it applicable to low-cost, small-scale projection systems and other domains.

optics,physics, applied
Holographic near-eye display system based on double-convergence light Gerchberg-Saxton algorithm.

Peng Sun,Shengqian Chang,Siqi Liu,Xiao Tao,Chang Wang,Zhenrong Zheng

DOI: https://doi.org/10.1364/OE.26.010140

IF: 3.8

2018-01-01

Optics Express

Abstract:In this paper. a method is proposed to implement noises reduced three-dimensional (3D) holographic near-eye display by phase-only computer-generated hologram (CGH). The CGH is calculated from a double-convergence light Gerchberg-Saxton (GS) algorithm, in which the phases of two virtual convergence lights are introduced into GS algorithm simultaneously. The first phase of convergence light is a replacement of random phase as the iterative initial value and the second phase of convergence light will modulate the phase distribution calculated by GS algorithm. Both simulations and experiments are carried out to verify the feasibility of the proposed method. The results indicate that this method can effectively reduce the noises in the reconstruction. Field of view (FOV) of the reconstructed image reaches 40 degrees and experimental light path in the 4-f system is shortened. As for 3D experiments, the results demonstrate that the proposed algorithm can present 3D images with 180cm zooming range and continuous depth cues. This method may provide a promising solution in future 3D augmented reality (AR) realization. (C) 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Computer-generated holographic near-eye display system based on LCoS phase only modulator

Peng Sun,Shengqian Chang,Siman Zhang,Ting Xie,Huaye Li,Siqi Liu,Chang Wang,Xiao Tao,Zhenrong Zheng

DOI: https://doi.org/10.1117/12.2271730

2017-01-01

Abstract:Augmented reality (AR) technology has been applied in various areas, such as large-scale manufacturing, national defense, healthcare, movie and mass media and so on. An important way to realize AR display is using computer-generated hologram (CGH), which is accompanied by low image quality and heavy computing defects. Meanwhile, the diffraction of Liquid Crystal on Silicon (LCoS) has a negative effect on image quality. In this paper, a modified algorithm based on traditional Gerchberg-Saxton (GS) algorithm was proposed to improve the image quality, and new method to establish experimental system was used to broaden field of view (FOV). In the experiment, undesired zero-order diffracted light was eliminated and high definition 2D image was acquired with FOV broadened to 36.1 degree. We have also done some pilot research in 3D reconstruction with tomography algorithm based on Fresnel diffraction. With the same experimental system, experimental results demonstrate the feasibility of 3D reconstruction. These modifications are effective and efficient, and may provide a better solution in AR realization.
Large Field-of-view Lensless Holographic Dynamic Projection System with Uniform Illumination and U-net Acceleration

Pin Su,Junrong Wang,Chao Cai,Jianshe Ma,Qiaofeng Tan

DOI: https://doi.org/10.1016/j.optlaseng.2022.107106

IF: 5.666

2022-01-01

Optics and Lasers in Engineering

Abstract:Compared to traditional projection systems, a diffraction-based lensless projection system offers the advantages of a compact size with high design freedom. However, the existing iterative algorithms for computer-generated holograms (CGHs) in large field-of-view diffraction systems are not mature, and are characterised by long process -ing times, making them unsuitable for application in dynamic displays. Furthermore, most large-scale diffraction algorithms neglect the intensity and phase nonuniformity of the illuminated spherical beam. Herein, we develop a large field-of-view laser projection system with U-net acceleration. Our system has a diffractive optical element (DOE) inserted between the digital micromirror device (DMD) and the screen to reduce the intensity and phase nonuniformity introduced by the spherical wave. Structural parameters of the system are also optimized, then am-plitude CGHs are loaded on the DMD sequentially. A divergent laser beam illuminates the DMD and is converted to a uniform beam by the DOE before it reaches the screen. The Gerchberg-Saxton (GS) iterative algorithm based on the scaled angular spectrum method (SASM-GS) is used to generate the amplitude CGH. However, as this process of computation includes three forward and three backward diffractions, it is time-consuming. Therefore, CGHs generated by the SASM-GS algorithm were used as labels, and a U-net model was trained using the MNIST database. Simulations revealed that the trained U-net model can accurately generate amplitude CGHs and has a certain degree of versatility. Notably, the hologram generation time of the U-net model was one-tenth that of the SASM-GS algorithm. Finally, we performed experiments to verify the efficacy of the U-net model, demonstrating the feasibility of the proposed dynamic lensless projection system.
Real-time High-Quality Computer-Generated Hologram Using Complex-Valued Convolutional Neural Network

Chongli Zhong,Xinzhu Sang,Binbin Yan,Hui Li,Duo Chen,Xiujuan Qin,Shuo Chen,Xiaoqian Ye

DOI: https://doi.org/10.1109/tvcg.2023.3239670

IF: 5.2

2023-01-01

IEEE Transactions on Visualization and Computer Graphics

Abstract:Holographic displays are ideal display technologies for virtual and augmented reality because all visual cues are provided. However, real-time high-quality holographic displays are difficult to achieve because the generation of high-quality computer-generated hologram (CGH) is inefficient in existing algorithms. Here, complex-valued convolutional neural network (CCNN) is proposed for phase-only CGH generation. The CCNN-CGH architecture is effective with a simple network structure based on the character design of complex amplitude. A holographic display prototype is set up for optical reconstruction. Experiments verify that state-of-the-art performance is achieved in terms of quality and generation speed in existing end-to-end neural holography methods using the ideal wave propagation model. The generation speed is three times faster than HoloNet and one-sixth faster than Holo-encoder, and the Peak Signal to Noise Ratio (PSNR) is increased by 3 dB and 9 dB, respectively. Real-time high-quality CGHs are generated in 1920×1072 and 3840×2160 resolutions for dynamic holographic displays.

computer science, software engineering
Generating Real-Scene Hologram Through Light Field Imaging and Deep Learning

Rui Wang,Lingyu Ai,Yinghui Wang,Xiaolong Zhu,Yuqing Ni,Myungjin Cho

DOI: https://doi.org/10.1088/2040-8986/ad4b3a

IF: 2.1

2024-01-01

Journal of Optics

Abstract:Addressing the challenge of acquiring holograms from real-world scenes, this study introduces a novel approach leveraging light field cameras to capture light field data, which is subsequently transformed into authentic scene holograms. This methodology integrates light field imaging technology with a pre-trained deep neural network. To compensate for the limitations inherent in camera hardware, a super-resolution algorithm is employed. The conversion of light field information into RGB-D data facilitates its input into the deep neural network, enabling the inference of corresponding real-world scene holograms. Empirical evidence demonstrates that the system is capable of inferring high-resolution (1920 x 1080) real-world scene holograms within a timeframe of 5 s, utilizing hardware comprising an NVIDIA RTX 3060.
Unsupervised Fourier-inspired neural network for real-time and high-fidelity computer-generated holography

Zhenxing Dong,Chao Xu,Yuye Ling,Yan Li,Yikai Su

DOI: https://doi.org/10.1364/ol.477630

IF: 3.6

2022-01-01

Optics Letters

Abstract:Learning-based computer-generated holography (CGH) algorithms appear as novel alternatives to generate phase-only holograms. However, most existing learning-based approaches underperform their iterative peers regarding display quality. Here, we recognize that current convolutional neural networks have difficulty learning cross-domain tasks due to the limited receptive field. In order to overcome this limitation, we propose a Fourier-inspired neural module, which can be easily integrated into various CGH frameworks and significantly enhance the quality of reconstructed images. By explicitly leveraging Fourier transforms within the neural network architecture, the mesoscopic information within the phase-only hologram can be more handily extracted. Both simulation and experiment were performed to showcase its capability. By incorporating it into U-Net and HoloNet, the peak signal-to-noise ratio of reconstructed images is measured at 29.16 dB and 33.50 dB during the simulation, which is 4.97 dB and 1.52 dB higher than those by the baseline U-Net and HoloNet, respectively. Similar trends are observed in the experimental results. We also experimentally demonstrated that U-Net and HoloNet with the proposed module can generate a monochromatic 1080p hologram in 0.015 s and 0.020 s, respectively.
High-speed Computer-Generated Holography Using an Autoencoder-Based Deep Neural Network

Jiachen Wu,Kexuan Liu,Xiaomeng Sui,Liangcai Cao

DOI: https://doi.org/10.1364/ol.425485

2021-01-01

Abstract:Learning-based computer-generated holography (CGH) provides a rapid hologram generation approach for holographic displays. Supervised training requires a large-scale dataset with target images and corresponding holograms. We propose an autoencoder-based neural network (holoencoder) for phase-only hologram generation. Physical diffraction propagation was incorporated into the autoencoder’s decoding part. The holoencoder can automatically learn the latent encodings of phase-only holograms in an unsupervised manner. The proposed holoencoder was able to generate high-fidelity 4K resolution holograms in 0.15 s. The reconstruction results validate the good generalizability of the holoencoder, and the experiments show fewer speckles in the reconstructed image compared with the existing CGH algorithms.
Unsupervised deep neural network for fast computer-generated holography with continuous depth control

Yuhang Zheng,Chenhang Shen,Zhu Wang,Yifei Xie,Weilong Zhou,Zichun Le

DOI: https://doi.org/10.1016/j.optlaseng.2024.108310

IF: 5.666

2024-05-16

Optics and Lasers in Engineering

Abstract:In computer-generated holography, leveraging deep learning for holographic display with continuous depth poses challenges in balancing depth cues and runtime. Sufficient data are typically not available for existing deep neural networks based on point-based datasets. To address this, we propose an unsupervised learning neural network named Polygon-based Holo, based on an efficient diffraction model that enables calculations from a reference plane to tilted plane to generate the target field wavefront. Polygon-based Holo is trained using the proposed diffraction model through an unsupervised process, employing mid-holograms superimposed with a small number of tilted planes. Polygon-based Holo can accurately predict wavefronts from input RGB-D images with continuous depth, displaying exceptional generalization across applications. The numerical simulations and optical experiments validate our approach, emphasizing its feasibility and potential for facilitating arbitrary depth control and dynamic display of holography.

optics
High-fidelity, Model-Driven Deep Learning Network for Phase-Only Computer-Generated Holography (conference Presentation)

Liangcai Cao,Kexuan Liu,Jiachen Wu,Zehao He

DOI: https://doi.org/10.1117/12.2655512

2023-01-01

Abstract:The combination of Computer-Generated Holography (CGH) and deep learning has opened the possibility to generate both real-time and high-quality holograms. However, the widely-used data-driven deep learning method faces the problem of the large number of labeled training datasets generated by traditional algorithms, such as Gerchberg–Saxton (GS) iterative algorithm. It always takes a long time and limits the training performance of the network. In this work, we propose a model-driven neural network for high-fidelity Phase-Only Hologram (POH) generation. The Fresnel diffraction process is introduced as the physical model, which makes the network can automatically learn the latent encodings of POHs in an unsupervised way. Furthermore, the sub-pixel convolution upsampling method effectively improves the reconstruction quality. Once the training is completed, the POH of any two-dimensional image can be quickly generated. The calculation time is one to two orders of magnitude faster.
Research on convolutional neural network to realize high-quality dynamic holographic display

Yanjun Fu,Hailong Li,Baiheng Ma,Zhanjun Yan

DOI: https://doi.org/10.1016/j.optlastec.2024.111997

IF: 4.939

2024-10-26

Optics & Laser Technology

Abstract:There are still some challenges in the computation speed and reproduction quality of holographic display. In this paper, a method is proposed to realize a high-quality 3D holographic display by using a complex value convolutional neural network. Firstly, the layer-based method is used to cut and layer the object. The amplitude and phase of each layer of the object are calculated by a complex value network at the same time. Then, the output of the complex value type of the network is converted into a phase-only hologram, and each layer of the hologram is superimposed to reproduce the complete 3D object. In addition, a novel color convolutional neural network is used to generate R, G and B three-color holograms simultaneously, reducing the time required for three-channel recalculation of holograms. The proposed method only takes 16 ms to generate a hologram from single-layer object, the average peak signal-to-noise ratio of the reconstructed image is 28 dB, and the structural similarity exceeds 0.99, which realizes a high-quality dynamic holographic display. The experimental results confirm the feasibility of the proposed method and provide a new method for real-time holographic display of 3D object.

optics,physics, applied
Generating Multi‐Depth 3D Holograms Using a Fully Convolutional Neural Network

Xingpeng Yan,Xinlei Liu,Jiaqi Li,Yanan Zhang,Hebin Chang,Tao Jing,Hairong Hu,Qiang Qu,Xi Wang,Xiaoyu Jiang

DOI: https://doi.org/10.1002/advs.202308886

IF: 15.1

2024-05-10

Advanced Science

Abstract:This work introduces the forward–backward‐diffraction framework for computing multi‐depth diffraction fields and the layer‐by‐layer replacement method for handling occlusion relationships. When combined with a fully convolutional neural network, it generates multi‐depth holograms with excellent depth focusing and corrects occlusion relationships. The reconstructed scene exhibits minimal speckle noise and few edge artifacts. Efficiently generating 3D holograms is one of the most challenging research topics in the field of holography. This work introduces a method for generating multi‐depth phase‐only holograms using a fully convolutional neural network (FCN). The method primarily involves a forward–backward‐diffraction framework to compute multi‐depth diffraction fields, along with a layer‐by‐layer replacement method (L2RM) to handle occlusion relationships. The diffraction fields computed by the former are fed into the carefully designed FCN, which leverages its powerful non‐linear fitting capability to generate multi‐depth holograms of 3D scenes. The latter can smooth the boundaries of different layers in scene reconstruction by complementing information of occluded objects, thus enhancing the reconstruction quality of holograms. The proposed method can generate a multi‐depth 3D hologram with a PSNR of 31.8 dB in just 90 ms for a resolution of 2160 × 3840 on the NVIDIA Tesla A100 40G tensor core GPU. Additionally, numerical and experimental results indicate that the generated holograms accurately reconstruct clear 3D scenes with correct occlusion relationships and provide excellent depth focusing.

materials science, multidisciplinary,nanoscience & nanotechnology,chemistry
Magnification of Optical Image in Holography Projection Using Lensless Fresnel Holography

Jianshe Ma,Ping Su,Feipeng Xia,Zhenbo Ren,Tong Liu

DOI: https://doi.org/10.1117/1.oe.51.8.085801

IF: 1.3

2012-01-01

Optical Engineering

Abstract:A digital micro-mirror device (DMD) acting as a real-time hologram is an emerging technology in dynamic holographic projection. This paper presents a lensless image magnification method in DMD holography by using a Fresnel hologram. By analyzing the diffraction order distribution in the image plane of a hologram produced by DMD, we find the factors that limit the size of the magnified image. We perform a lensless magnification experiment that shows good magnified images in accordance with the numerical results. Finally, we discuss methods to eliminate longitudinal error and chromatic aberration in three-dimensional (3-D) and color projection, respectively, and present a 3-D image reconstruction result that shows lensless magnification of a 3-D image without distortion. It is believed that this technique can be used in future real-time holographic projection based on digital light processing technology. (C) 2012 Society of Photo-Optical Instrumentation Engineers (SPIE). [DOI: 10.1117/1.OE.51.8.085801]
Deep-learning-aided multi-focal hologram generation

Zhelin Qu,Hao Jiang,Kunpeng Wang,Yunxia Liu,Yang Yang,Chao Feng,Xian Zhao,Jun-Lei Wang

DOI: https://doi.org/10.1016/j.optlastec.2024.112056

IF: 4.939

2024-11-10

Optics & Laser Technology

Abstract:Due to its superior reconfigurability and multifunctionality, computer-generated holograms (CGH) have found extensive utilization in research of light field modulation. Compared with successive light fields such as images, point light fields with multiple foci place higher demands on hologram quality. Moreover, the increasing complexity of the target light field and spatial resolution of holographic displays lead to a corresponding surge in the computation demand for generating high-quality holograms as well as the increasing difficulty to achieve real-time hologram manipulation without compromising the quality. In this study, we demonstrate a deep learning (DL) aided CGH method capable of high fidelity and rapid generation of holograms for multi-foci light field applications by prediction of intrinsic weighting and phase parameter pairs w,θ . Unlike previous DL-based methods that primarily focus on successive light field generation in fixed optical configuration, this approach not only facilitate the initial generation of rapid, high-performance and versatile multi-foci light field in any adjustable optical system, but also allows for subsequent rapid iterative refinements. A lightweight convolutional neural network (CNN) framework is devised to learn the mapping between input light field distributions and w,θ parameters with the linearly corrected root-mean-square error as loss in an unsupervised manner. Our approach stands out in both a non-iterative algorithm with enhanced quality and an iterative algorithm with faster computation speed, adapting to the factual demands while effectively mitigating undesired DL-induced results. The effectiveness of this method is validated through both numerical simulations and experiment demonstrations.

optics,physics, applied
Towards real-time photorealistic 3D holography with deep neural networks

Liang Shi,Beichen Li,Changil Kim,Petr Kellnhofer,Wojciech Matusik

DOI: https://doi.org/10.1038/s41586-020-03152-0

IF: 64.8

2021-03-10

Nature

Abstract:The ability to present three-dimensional (3D) scenes with continuous depth sensation has a profound impact on virtual and augmented reality, human–computer interaction, education and training. Computer-generated holography (CGH) enables high-spatio-angular-resolution 3D projection via numerical simulation of diffraction and interference<a href="/articles/s41586-020-03152-0#ref-CR1">1</a>. Yet, existing physically based methods fail to produce holograms with both per-pixel focal control and accurate occlusion<a href="/articles/s41586-020-03152-0#ref-CR2">2</a>,<a href="/articles/s41586-020-03152-0#ref-CR3">3</a>. The computationally taxing Fresnel diffraction simulation further places an explicit trade-off between image quality and runtime, making dynamic holography impractical<a href="/articles/s41586-020-03152-0#ref-CR4">4</a>. Here we demonstrate a deep-learning-based CGH pipeline capable of synthesizing a photorealistic colour 3D hologram from a single RGB-depth image in real time. Our convolutional neural network (CNN) is extremely memory efficient (below 620 kilobytes) and runs at 60 hertz for a resolution of 1,920 × 1,080 pixels on a single consumer-grade graphics processing unit. Leveraging low-power on-device artificial intelligence acceleration chips, our CNN also runs interactively on mobile (iPhone 11 Pro at 1.1 hertz) and edge (Google Edge TPU at 2.0 hertz) devices, promising real-time performance in future-generation virtual and augmented-reality mobile headsets. We enable this pipeline by introducing a large-scale CGH dataset (MIT-CGH-4K) with 4,000 pairs of RGB-depth images and corresponding 3D holograms. Our CNN is trained with differentiable wave-based loss functions<a href="/articles/s41586-020-03152-0#ref-CR5">5</a> and physically approximates Fresnel diffraction. With an anti-aliasing phase-only encoding method, we experimentally demonstrate speckle-free, natural-looking, high-resolution 3D holograms. Our learning-based approach and the Fresnel hologram dataset will help to unlock the full potential of holography and enable applications in metasurface design<a href="/articles/s41586-020-03152-0#ref-CR6">6</a>,<a href="/articles/s41586-020-03152-0#ref-CR7">7</a>, optical and acoustic tweezer-based microscopic manipulation<a href="#ref-CR8">8</a>,<a href="#ref-CR9">9</a>,<a href="/articles/s41586-020-03152-0#ref-CR10">10</a>, holographic microscopy<a href="/articles/s41586-020-03152-0#ref-CR11">11</a> and single-exposure volumetric 3D printing<a href="/articles/s41586-020-03152-0#ref-CR12">12</a>,<a href="/articles/s41586-020-03152-0#ref-CR13">13</a>.

multidisciplinary sciences
High Bandwidth-Utilization Digital Holographic Reconstruction Using an Untrained Neural Network

Zhuoshi Li,Yuanyuan Chen,Jiasong Sun,Yanbo Jin,Qian Shen,Peng Gao,Qian Chen,Chao Zuo

DOI: https://doi.org/10.3390/app122010656

2022-01-01

Applied Sciences

Abstract:Slightly off-axis digital holographic microscopy (DHM) is the extension of digital holography imaging technology toward high-throughput modern optical imaging technology. However, it is difficult for the method based on the conventional linear Fourier domain filtering to solve the imaging artifacts caused by the spectral aliasing problem. In this article, we propose a novel high-accuracy, artifacts-free, single-frame, digital holographic phase demodulation scheme for low-carrier-frequency holograms, which incorporates the physical model into a conventional deep neural network (DNN) without training beforehand based on a massive dataset. Although the conventional end-to-end deep learning (DL) method can achieve high-accuracy phase recovery directly from a single-frame hologram, the massive datasets and ground truth collection can be prohibitively laborious and time-consuming. Our method recognizes such a low-carrier frequency fringe demodulation process as a nonlinear optimization problem, which can reconstruct the artifact-free phase details gradually from a single-frame hologram. The phase resolution target and simulation experiment results quantitatively demonstrate that the proposed method possesses better artifact suppression and high-resolution imaging capabilities than the physical methods. In addition, the live-cell experiment also indicates the practicality of the technique in biological research.
Nonlinear Holographic Image Inpainting Via Deep Learning

Gan Wu,Huan Chen,Xuhui Sun,Yizheng Yao,Tong Wang,Yibing Ma,Chenglong Wang,Bing Gao,Hao Wu,Ronger Lu,Chao Zhang,Yiqiang Qin

DOI: https://doi.org/10.1117/1.oe.62.7.075101

IF: 1.3

2023-01-01

Optical Engineering

Abstract:Nonlinear photonic crystals can be employed to generate holographic images in nonlinear optical processes. However, due to the limit of binary structure, the process of holographic imaging will lose part of the amplitude information and cause image hollowing, thus the imaging quality is reduced. Generative adversarial network, a deep learning network based on game theory, is used to restore images. The results of restoration show high similarity to the original images, effectively weakening the effect of image hollowing, suppressing the diffraction effect, and restoring grayscale values. This image post-processing approach completes the field of application of nonlinear holographic imaging, which is useful for non-visible source imaging, crosstalk avoidance, optical encryption, and so on. (c) 2023 Society of Photo- Optical Instrumentation Engineers (SPIE)
Liquid lens based holographic camera for real 3D scene hologram acquisition using end-to-end physical model-driven network

Di Wang,Zhao-Song Li,Yi Zheng,You-Ran Zhao,Chao Liu,Jin-Bo Xu,Yi-Wei Zheng,Qian Huang,Chen-Liang Chang,Da-Wei Zhang,Song-Lin Zhuang,Qiong-Hua Wang

DOI: https://doi.org/10.1038/s41377-024-01410-8

2024-02-29

Abstract:Abstract With the development of artificial intelligence, neural network provides unique opportunities for holography, such as high fidelity and dynamic calculation. How to obtain real 3D scene and generate high fidelity hologram in real time is an urgent problem. Here, we propose a liquid lens based holographic camera for real 3D scene hologram acquisition using an end-to-end physical model-driven network (EEPMD-Net). As the core component of the liquid camera, the first 10 mm large aperture electrowetting-based liquid lens is proposed by using specially fabricated solution. The design of the liquid camera ensures that the multi-layers of the real 3D scene can be obtained quickly and with great imaging performance. The EEPMD-Net takes the information of real 3D scene as the input, and uses two new structures of encoder and decoder networks to realize low-noise phase generation. By comparing the intensity information between the reconstructed image after depth fusion and the target scene, the composite loss function is constructed for phase optimization, and the high-fidelity training of hologram with true depth of the 3D scene is realized for the first time. The holographic camera achieves the high-fidelity and fast generation of the hologram of the real 3D scene, and the reconstructed experiment proves that the holographic image has the advantage of low noise. The proposed holographic camera is unique and can be used in 3D display, measurement, encryption and other fields.

optics
Large-Field Holographic Projection System Based On Deep-Learning Acceleration Calculation

Chao Cai,Ping Su,Jianshe Ma

DOI: https://doi.org/10.1117/12.2573590

2020-01-01

Abstract:The large field of view projection system based on the principle of diffraction has a simple optical system, but it often consumes a lot of time when calculating the hologram using the traditional iterative algorithm, which limits the application range of the system. In this paper, we use deep learning to accelerate the calculation of holograms, and experiment results proves that the holograms calculated by U-net have good performance.

Lensless Magnified Holographic Projection Based on an Unsupervised Neural Network Technology

All-optical Diffractive Neural Networked Terahertz Hologram.

Lensless holographic dynamic projection system based on weakly supervised learning

Holographic near-eye display system based on double-convergence light Gerchberg-Saxton algorithm.

Computer-generated holographic near-eye display system based on LCoS phase only modulator

Large Field-of-view Lensless Holographic Dynamic Projection System with Uniform Illumination and U-net Acceleration

Real-time High-Quality Computer-Generated Hologram Using Complex-Valued Convolutional Neural Network

Generating Real-Scene Hologram Through Light Field Imaging and Deep Learning

Unsupervised Fourier-inspired neural network for real-time and high-fidelity computer-generated holography

High-speed Computer-Generated Holography Using an Autoencoder-Based Deep Neural Network

Unsupervised deep neural network for fast computer-generated holography with continuous depth control

High-fidelity, Model-Driven Deep Learning Network for Phase-Only Computer-Generated Holography (conference Presentation)

Research on convolutional neural network to realize high-quality dynamic holographic display

Generating Multi‐Depth 3D Holograms Using a Fully Convolutional Neural Network

Magnification of Optical Image in Holography Projection Using Lensless Fresnel Holography

Deep-learning-aided multi-focal hologram generation

Towards real-time photorealistic 3D holography with deep neural networks

High Bandwidth-Utilization Digital Holographic Reconstruction Using an Untrained Neural Network

Nonlinear Holographic Image Inpainting Via Deep Learning

Liquid lens based holographic camera for real 3D scene hologram acquisition using end-to-end physical model-driven network

Large-Field Holographic Projection System Based On Deep-Learning Acceleration Calculation