The popularity of deep learning has boosted computer-generated holography(CGH)as a vibrant research field,particularly physics-driven unsupervised learning.Nevertheless,present unsupervised CGH models have not yet exp...The popularity of deep learning has boosted computer-generated holography(CGH)as a vibrant research field,particularly physics-driven unsupervised learning.Nevertheless,present unsupervised CGH models have not yet explored the potential of generating full-color 3D holograms through a unified framework.In this study,we propose a lightweight multiwavelength network model capable of high-fidelity and efficient full-color hologram generation in both 2D and 3D display,called IncepHoloRGB.The high-speed simultaneous generation of RGB holograms at 191 frames per second(FPS)is based on Inception sampling blocks and multi-wavelength propagation module integrated with depth-traced superimposition,achieving an average structural similarity(SSIM)of 0.88 and peak signal-to-noise ratio(PSNR)of 29.00 on the DIV2K test set in reconstruction.Full-color reconstruction of numerical simulations and optical experiments shows that IncepHoloRGB is versatile to diverse scenarios and can obtain authentic full-color holographic 3D display within a unified network model,paving the way for applications towards real-time dynamic naked-eye 3D display,virtual and augmented reality(VR/AR)systems.展开更多
Deep learning offers a novel opportunity to achieve both high-quality and high-speed computer-generated holography(CGH).Current data-driven deep learning algorithms face the challenge that the labeled training dataset...Deep learning offers a novel opportunity to achieve both high-quality and high-speed computer-generated holography(CGH).Current data-driven deep learning algorithms face the challenge that the labeled training datasets limit the training performance and generalization.The model-driven deep learning introduces the diffraction model into the neural network.It eliminates the need for the labeled training dataset and has been extensively applied to hologram generation.However,the existing model-driven deep learning algorithms face the problem of insufficient constraints.In this study,we propose a model-driven neural network capable of high-fidelity 4K computer-generated hologram generation,called 4K Diffraction Model-driven Network(4K-DMDNet).The constraint of the reconstructed images in the frequency domain is strengthened.And a network structure that combines the residual method and sub-pixel convolution method is built,which effectively enhances the fitting ability of the network for inverse problems.The generalization of the 4K-DMDNet is demonstrated with binary,grayscale and 3D images.High-quality full-color optical reconstructions of the 4K holograms have been achieved at the wavelengths of 450 nm,520 nm,and 638 nm.展开更多
We present a novel method for scale-invariant 3D face recognition by integrating computer-generated holography with the Mellin transform.This approach leverages the scale-invariance property of the Mellin transform to...We present a novel method for scale-invariant 3D face recognition by integrating computer-generated holography with the Mellin transform.This approach leverages the scale-invariance property of the Mellin transform to address challenges related to variations in 3D facial sizes during recognition.By applying the Mellin transform to computer-generated holograms and performing correlation between them,which,to the best of our knowledge,is being done for the first time,we have developed a robust recognition framework capable of managing significant scale variations without compromising recognition accuracy.Digital holograms of 3D faces are generated from a face database,and the Mellin transform is employed to enable robust recognition across scale factors ranging from 0.4 to 2.0.Within this range,the method achieves 100%recognition accuracy,as confirmed by both simulation-based and hybrid optical/digital experimental validations.Numerical calculations demonstrate that our method significantly enhances the accuracy and reliability of 3D face recognition,as evidenced by the sharp correlation peaks and higher peak-to-noise ratio(PNR)values than that of using conventional holograms without the Mellin transform.Additionally,the hybrid optical/digital joint transform correlation hardware further validates the method's effectiveness,demonstrating its capability to accurately identify and distinguish 3D faces at various scales.This work provides a promising solution for advanced biometric systems,especially for those which require 3D scale-invariant recognition.展开更多
This Paper studies a reflection-type 2-D computer-generated holographic phase grating used in laser coherent imaging Systems as beam shaping elements. In the applications the diffractive gratings must have high diffra...This Paper studies a reflection-type 2-D computer-generated holographic phase grating used in laser coherent imaging Systems as beam shaping elements. In the applications the diffractive gratings must have high diffractionefficiency and produce uniform intensities in the diffracted other desired. In this paper we discuss in some detail theuse of conjugation gratings applied mainly to laser coherent imaging systems with detector arrays to create 2-D gratingstructures that produce multi-fold diffracted beams.展开更多
We propose a Phong shading approximation,which gives the amplitude of each point inside the triangle through linear interpolation within the framework of self-similarity segmentation and affine transformation in polyg...We propose a Phong shading approximation,which gives the amplitude of each point inside the triangle through linear interpolation within the framework of self-similarity segmentation and affine transformation in polygon-based computer-generated holography.Shading is important as it reflects the geometric properties of the objects.To accurately represent the geometric properties of objects in three-dimensional space,the method involves calculating the amplitude distribution on each triangle and maintaining a complete analytical framework,with the edges of the reconstructed polygons nearly unobservable.Numerical simulations and optical reconstructions demonstrate that the proposed method successfully addresses the issue of edge discontinuity on polygonal surfaces.展开更多
Incoherent digital holography has attracted significant attention due to its advantages in threedimensional(3D)imaging under low spatial coherence conditions,such as easy access to light sources and reduced speckle no...Incoherent digital holography has attracted significant attention due to its advantages in threedimensional(3D)imaging under low spatial coherence conditions,such as easy access to light sources and reduced speckle noise.However,interlayer crosstalk during the reconstruction process leads to a substantial reduction in reconstruction fidelity.Furthermore,existing deconvolutionand deep-learning-based reconstruction algorithms face limitations in terms of effectiveness and generalization.To address these challenges,we propose a compressive incoherent digital holography(CIDH)approach for 3D imaging.In CIDH,a point spread hologram sequence with a high signal-to-noise ratio is initially obtained using a customized computergenerated holography method for dual-channel forward data acquisition.For scene reconstruction,a compressed sensing-based two-step iterative shrinkage/thresholding algorithm is employed to achieve high-fidelity 3D scene retrieval.The combined optimization demonstrates exceptional performance in suppressing interlayer crosstalk and enhancing reconstruction fidelity.In simulations,crosstalk was effectively suppressed across 10 depth layers.In experiments,successful suppression was achieved for both a five-layer transmissive object and a two-layer reflective 3D object,resulting in significantly improved reconstruction accuracy.The proposed framework shows great potential for applications in various incoherent source-illuminated and fluorescent 3D imaging.展开更多
A phase-only computer-generated holography（CGH） calculation method for stereoscopic holography is proposed in this paper.The two-dimensional（2D） perspective projection views of the three-dimensional（3D） object a...A phase-only computer-generated holography（CGH） calculation method for stereoscopic holography is proposed in this paper.The two-dimensional（2D） perspective projection views of the three-dimensional（3D） object are generated by the computer graphics rendering techniques.Based on these views,a phase-only hologram is calculated by using the Gerchberg–Saxton（GS） iterative algorithm.Comparing with the non-iterative algorithm in the conventional stereoscopic holography,the proposed method improves the holographic image quality,especially for the phase-only hologram encoded from the complex distribution.Both simulation and optical experiment results demonstrate that our proposed method can give higher quality reconstruction comparing with the traditional method.展开更多
With the explosive growth of mathematical optimization and computing hardware,deep neural networks(DNN)have become tremendously powerful tools to solve many challenging problems in various fields,ranging from decision...With the explosive growth of mathematical optimization and computing hardware,deep neural networks(DNN)have become tremendously powerful tools to solve many challenging problems in various fields,ranging from decision making to computational imaging and holography.In this manuscript,I focus on the prosperous interactions between DNN and holography.On the one hand,DNN has been demonstrated to be in particular proficient for holographic reconstruction and computer-generated holography almost in every aspect.On the other hand,holography is an enabling tool for the optical implementation of DNN the other way around owing to the capability of interconnection and light speed processing in parallel.The purpose of this article is to give a comprehensive literature review on the recent progress of deep holography,an emerging interdisciplinary research field that is mutually inspired by holography and DNN.I first give a brief overview of the basic theory and architectures of DNN,and then discuss some of the most important progresses of deep holography.I hope that the present unified exposition will stimulate further development in this promising and exciting field of research.展开更多
基金supports from National Natural Science Foundation of China(Grant No.62205117,52275429)National Key Research and Development Program of China(Grant No.2021YFF0502700)+2 种基金Young Elite Scientists Sponsorship Program by CAST(Grant No.2022QNRC001)West Light Foundation of the Chinese Academy of Sciences(Grant No.xbzg-zdsys-202206)Hubei Natural Science Foundation Innovative Research Group Project(2024AFA025).
文摘The popularity of deep learning has boosted computer-generated holography(CGH)as a vibrant research field,particularly physics-driven unsupervised learning.Nevertheless,present unsupervised CGH models have not yet explored the potential of generating full-color 3D holograms through a unified framework.In this study,we propose a lightweight multiwavelength network model capable of high-fidelity and efficient full-color hologram generation in both 2D and 3D display,called IncepHoloRGB.The high-speed simultaneous generation of RGB holograms at 191 frames per second(FPS)is based on Inception sampling blocks and multi-wavelength propagation module integrated with depth-traced superimposition,achieving an average structural similarity(SSIM)of 0.88 and peak signal-to-noise ratio(PSNR)of 29.00 on the DIV2K test set in reconstruction.Full-color reconstruction of numerical simulations and optical experiments shows that IncepHoloRGB is versatile to diverse scenarios and can obtain authentic full-color holographic 3D display within a unified network model,paving the way for applications towards real-time dynamic naked-eye 3D display,virtual and augmented reality(VR/AR)systems.
基金We are grateful for financial supports from National Natural Science Foundation of China(62035003,61775117)China Postdoctoral Science Foundation(BX2021140)Tsinghua University Initiative Scientific Research Program(20193080075).
文摘Deep learning offers a novel opportunity to achieve both high-quality and high-speed computer-generated holography(CGH).Current data-driven deep learning algorithms face the challenge that the labeled training datasets limit the training performance and generalization.The model-driven deep learning introduces the diffraction model into the neural network.It eliminates the need for the labeled training dataset and has been extensively applied to hologram generation.However,the existing model-driven deep learning algorithms face the problem of insufficient constraints.In this study,we propose a model-driven neural network capable of high-fidelity 4K computer-generated hologram generation,called 4K Diffraction Model-driven Network(4K-DMDNet).The constraint of the reconstructed images in the frequency domain is strengthened.And a network structure that combines the residual method and sub-pixel convolution method is built,which effectively enhances the fitting ability of the network for inverse problems.The generalization of the 4K-DMDNet is demonstrated with binary,grayscale and 3D images.High-quality full-color optical reconstructions of the 4K holograms have been achieved at the wavelengths of 450 nm,520 nm,and 638 nm.
基金financial supports from the National Natural Science Foundation of China(Grant No.6227511362405124).
文摘We present a novel method for scale-invariant 3D face recognition by integrating computer-generated holography with the Mellin transform.This approach leverages the scale-invariance property of the Mellin transform to address challenges related to variations in 3D facial sizes during recognition.By applying the Mellin transform to computer-generated holograms and performing correlation between them,which,to the best of our knowledge,is being done for the first time,we have developed a robust recognition framework capable of managing significant scale variations without compromising recognition accuracy.Digital holograms of 3D faces are generated from a face database,and the Mellin transform is employed to enable robust recognition across scale factors ranging from 0.4 to 2.0.Within this range,the method achieves 100%recognition accuracy,as confirmed by both simulation-based and hybrid optical/digital experimental validations.Numerical calculations demonstrate that our method significantly enhances the accuracy and reliability of 3D face recognition,as evidenced by the sharp correlation peaks and higher peak-to-noise ratio(PNR)values than that of using conventional holograms without the Mellin transform.Additionally,the hybrid optical/digital joint transform correlation hardware further validates the method's effectiveness,demonstrating its capability to accurately identify and distinguish 3D faces at various scales.This work provides a promising solution for advanced biometric systems,especially for those which require 3D scale-invariant recognition.
文摘This Paper studies a reflection-type 2-D computer-generated holographic phase grating used in laser coherent imaging Systems as beam shaping elements. In the applications the diffractive gratings must have high diffractionefficiency and produce uniform intensities in the diffracted other desired. In this paper we discuss in some detail theuse of conjugation gratings applied mainly to laser coherent imaging systems with detector arrays to create 2-D gratingstructures that produce multi-fold diffracted beams.
基金supported by the National Natural Science Foundation of China(Nos.62341124 and 62275113)the Yunnan Fundamental Research Projects(No.202201AT070030)。
文摘We propose a Phong shading approximation,which gives the amplitude of each point inside the triangle through linear interpolation within the framework of self-similarity segmentation and affine transformation in polygon-based computer-generated holography.Shading is important as it reflects the geometric properties of the objects.To accurately represent the geometric properties of objects in three-dimensional space,the method involves calculating the amplitude distribution on each triangle and maintaining a complete analytical framework,with the edges of the reconstructed polygons nearly unobservable.Numerical simulations and optical reconstructions demonstrate that the proposed method successfully addresses the issue of edge discontinuity on polygonal surfaces.
基金supported by the National Natural Science Foundation of China(Grant Nos.12325408,12274129,12374274,12274139,62175066,92150102,62475070,12474404,12471368,and 12304338)the Shanghai Municipal Education Commission(Grant No.2024AI01007).
文摘Incoherent digital holography has attracted significant attention due to its advantages in threedimensional(3D)imaging under low spatial coherence conditions,such as easy access to light sources and reduced speckle noise.However,interlayer crosstalk during the reconstruction process leads to a substantial reduction in reconstruction fidelity.Furthermore,existing deconvolutionand deep-learning-based reconstruction algorithms face limitations in terms of effectiveness and generalization.To address these challenges,we propose a compressive incoherent digital holography(CIDH)approach for 3D imaging.In CIDH,a point spread hologram sequence with a high signal-to-noise ratio is initially obtained using a customized computergenerated holography method for dual-channel forward data acquisition.For scene reconstruction,a compressed sensing-based two-step iterative shrinkage/thresholding algorithm is employed to achieve high-fidelity 3D scene retrieval.The combined optimization demonstrates exceptional performance in suppressing interlayer crosstalk and enhancing reconstruction fidelity.In simulations,crosstalk was effectively suppressed across 10 depth layers.In experiments,successful suppression was achieved for both a five-layer transmissive object and a two-layer reflective 3D object,resulting in significantly improved reconstruction accuracy.The proposed framework shows great potential for applications in various incoherent source-illuminated and fluorescent 3D imaging.
基金Project supported by the National Basic Research Program of China(Grant No.2013CB328803)the National High Technology Research and Development Program of China(Grant Nos.2013AA013904 and 2015AA016301)
文摘A phase-only computer-generated holography（CGH） calculation method for stereoscopic holography is proposed in this paper.The two-dimensional（2D） perspective projection views of the three-dimensional（3D） object are generated by the computer graphics rendering techniques.Based on these views,a phase-only hologram is calculated by using the Gerchberg–Saxton（GS） iterative algorithm.Comparing with the non-iterative algorithm in the conventional stereoscopic holography,the proposed method improves the holographic image quality,especially for the phase-only hologram encoded from the complex distribution.Both simulation and optical experiment results demonstrate that our proposed method can give higher quality reconstruction comparing with the traditional method.
基金supported by the National Natural Science Foundation of China(62061136005,61991452)the Sino-German Center(GZ1391)the Key Research Program of Frontier Sciences of the Chinese Academy of Sciences(QYZDB-SSW-JSC002).
文摘With the explosive growth of mathematical optimization and computing hardware,deep neural networks(DNN)have become tremendously powerful tools to solve many challenging problems in various fields,ranging from decision making to computational imaging and holography.In this manuscript,I focus on the prosperous interactions between DNN and holography.On the one hand,DNN has been demonstrated to be in particular proficient for holographic reconstruction and computer-generated holography almost in every aspect.On the other hand,holography is an enabling tool for the optical implementation of DNN the other way around owing to the capability of interconnection and light speed processing in parallel.The purpose of this article is to give a comprehensive literature review on the recent progress of deep holography,an emerging interdisciplinary research field that is mutually inspired by holography and DNN.I first give a brief overview of the basic theory and architectures of DNN,and then discuss some of the most important progresses of deep holography.I hope that the present unified exposition will stimulate further development in this promising and exciting field of research.
