Off-axis digital holographic reconstruction based on residual attention mechanism

MENG Mingwei; JIANG Xingguo

doi:null

您当前的位置：

首页 >

文章列表页 >

Off-axis digital holographic reconstruction based on residual attention mechanism

Optical Information Technology | 更新时间：2026-01-22

- Off-axis digital holographic reconstruction based on residual attention mechanism
- OPTICAL TECHNIQUE Vol. 52, Issue 1, Pages: 61-67(2026)
- 作者机构：
  
  1.智能感知与控制四川省重点实验室，四川宜宾 644005
  2.四川轻化工大学物理与电子工程学院，四川宜宾 644005
- 作者简介：
- 基金信息：
- DOI：
  CLC： O436
- Received：19 September 2025，
  
  Revised：2025-10-31，
  
  Published：20 January 2026
- 稿件说明：
移动端阅览
蒙明伟,蒋行国.基于残差注意力机制的离轴数字全息重建[J].光学技术,2026,52(01):61-67.

MENG Mingwei,JIANG Xingguo.Off-axis digital holographic reconstruction based on residual attention mechanism[J].OPTICAL TECHNIQUE,2026,52(01):61-67.
蒙明伟,蒋行国.基于残差注意力机制的离轴数字全息重建[J].光学技术,2026,52(01):61-67. DOI：

MENG Mingwei,JIANG Xingguo.Off-axis digital holographic reconstruction based on residual attention mechanism[J].OPTICAL TECHNIQUE,2026,52(01):61-67. DOI：

摘要

离轴数字全息重建具有重要应用价值，但重建方法仍较为有限，面临零级衍射干扰，计算复杂度高以及重建精度受限等问题。针对这些问题，论文提出一种基于残差注意力机制的离轴数字全息重建方法。首先采用优化后的U-Net模型作为网络架构，并使用模拟生成的离轴菲涅尔全息图进行训练。然后，通过改进U-Net模型中的编码器和解码器结构，以提升网络特征提取能力，从而能够实现高质量的离轴数字全息图重建。最终，为进一步验证该模型的抗噪声性能，保持网络结构和参数不变，使用含噪声的数据集进行重建，并对重建结果进行质量评估。实验结果表明，相较于最优对比算法，改进后的网络在数据集上的平均SSIM和PSNR分别提高了2.97%和15.52%，而在加噪数据集上平均SSIM和PSNR则分别提高了6.48%和19.04%。显著提升了离轴数字全息重建的质量和抗噪性能，为复杂光学成像问题提供了有效的深度学习解决方案。

Abstract

Off-axis digital holographic reconstruction holds significant practical value

yet existing methods remain limited due to challenges such as zero-order diffraction interference

high computational complexity

and restricted reconstruction accuracy. To address these issues

an off-axis digital holographic reconstruction method based on residual attention mechanisms is proposed. The study first employs an optimized U-Net model architecture and utilizes simulated off-axis Fresnel holograms for training. Subsequently

the encoder-decoder structures in the U-Net model are enhanced to improve feature extraction capabilities

enabling high-quality reconstruction of off-axis digital holograms. Finally

to validate the model's noise resistance

reconstruction was performed using noisy datasets while maintaining the original network structure and parameters

with quality assessments conducted on reconstructed results. Experimental results demonstrate that compared to the optimal reference algorithm

the improved network achieves 2.97% higher average SSIM and 15.52% better PSNR performance on standard datasets. Notably

it shows 6.48% and 19.04% improvements in average SSIM and PSNR across noisy datasets.The quality and noise reduction performance of off-axis digital holographic reconstruction are significantly improved

and an effective deep learning solution is provided for complex optical imaging problems.

关键词

Keywords

references

Zhong W . Application of artificial intelligence digital holography technology based on medical sensors in the development of medical image fusion [J ] . Measurement:Sensors , 2024 , 33 : 101146 .

张虎鸣 , 颜玢玢 , 仲崇力 , 等 . 基于HOE的全息显示双目视差构建方法 [J ] . 液晶与显示 , 2024 , 39 （ 07 ）: 892 — 900 .

Zhang Huming , Yan Panpan , Zhong Chongli , et al . A construction method of binocular parallax for holographic display based on HOE [J ] . Liquid Crystals And Displays , 2024 , 39 （ 07 ）: 892 — 900 .

Fosodeder P , Hubmer S , Ploier A , et al . Phase-contrast THz-CT for non-destructive testin [J ] . Optics Express , 2021 , 29 （ 10 ）: 15711 — 15723 .

张华 , 曹良才 , 金国藩 , 等 . 基于压缩感知算法的无透镜数字全息成像研究 [J ] . 激光与光电子学进展 , 2020 , 57 （ 08 ）: 9 — 19 .

Zhang Hua , Cao Liangcai , Jin Guopan , et al . Research on lensless digital holographic imaging based on compressed sensing algorithm [J ] . Laser & Optoelectronics Progress , 2020 ,57（ 08 ） 9 — 19 .

付泰 , 陈艾帅 , 胡磊 , 等 . 基于小波变换的全息图压缩编码比较研究 [J ] . 应用光学 , 2024 , 45 （ 01 ）: 99 — 106 .

Fu Tai , Chen Aishuai , Hu Lei , et al . A comparative study on hologram compression coding based on wavelet transform [J ] . Journal of Applied Optics , 2024 , 45 （ 01 ）: 99 — 106 .

Khalifa A F , Badr E . Deep learning for image segmentation: a focus on medical imaging [J ] . Comput. Mater. Contin , 2023 , 5 （ 1 ）: 1995 — 2024 .

Qiao C , Li D , Liu Y , et al . Rationalized deep learning super-resolution microscopy for sustained live imaging of rapid subcellular processes [J ] . Nature Biotechnology , 2023 , 41 （ 3 ）: 367 — 377 .

Liu J , Pan C , Yan W Q . Litter detection from digital images using deep learning [J ] . SN Computer Science , 2022 , 4 （ 2 ）: 134 .

Zhao X , Wang L , Zhang Y , et al . A review of convolutional neural networks in computer vision [J ] . Artificial Intelligence Review , 2024 , 57 （ 4 ）: 99 .

De Souza V L T , Marques B A D , Batagelo H C , et al . A review on generative adversarial networks for image generation [J ] . Computers & Graphics , 2023 , 114 : 13 — 25 .

Sherstinsky A . Fundamentals of recurrent neural network （RNN） and long short-term memory （LSTM） network [J ] . Physica D: Nonlinear Phenomena , 2020 , 404 : 132306 .

Wang K , Qian K , Di J , et al . Y4-Net: a deep learning solution to one-shot dual- wavelength digital holographic reconstruction [J ] . Optics Letters , 2020 , 45 （ 15 ）: 4220 — 4223 .

何晨 , 房鸿 , 张宁超 . 基于深度学习的单帧同轴数字全息重建方法 [J ] . 光子学报 , 2021 , 50 （ 12 ）: 275 — 284 .

He Chen , Fang Hong , Zhang Ningchao . Single-frame coaxial digital holographic reconstruction method based on deep learning [J ] . Acta Photonica Sinica , 2021 , 50 （ 12 ）: 275 — 284 .

浦健 , 桂进斌 , 张凯 . 基于深度学习的多尺度数字全息图重建 [J ] . 激光与光电子学进展 , 2022 , 59 （ 02 ）: 86 — 93 .

Pu Jian , Gui Jinbin , Zhang Kai . Multi-scale digital hologram reconstruction based on deep learning [J ] . Laser & Optoelectronics Progress , 2022 , 59 （ 02 ）: 86 — 93 .

成珂阳 , 李琦 . 深度学习用于连续太赫兹同轴数字全息重建 [J ] . 中国激光 , 2023 , 50 （ 19 ）: 246 — 256 .

Cheng Keyang , Li Qi . Deep learning is used for continuous terahertz coaxial digital holographic reconstruction [J ] . Chinese Journal Of Lasers , 2023 , 50 （ 19 ）: 246 — 256 .

刘俊彤 , 桂进斌 , 陈艾帅 , 等 . 基于深度学习的彩色全息图重建 [J ] . 激光与光电子学进展 , 2024 , 61 （ 08 ）: 86 — 92 .

Liu Juntong , Gui , Jinbin , Chen Aishuai , et al . Deep learning-based color hologram reconstruction [J ] . Laser & Optoelectronics Progress , 2024 , 61 （ 08 ）: 86 — 92 .

严利平 , 苗小喆 , 黄柳 . MScU-Net全息相位解包裹与像差同步求解 [J ] . 光学精密工程 , 2025 , 33 （ 15 ）: 2441 — 2454 .

Yan Liping , Miao xiaozhe , Huang Liu . MScU-Net holographic phase unwrapping and aberration synchronous solving [J ] . Optics and Precision Engineering , 2025 , 33 （ 15 ）: 2441 — 2454 .

胡先飞 , 桂进斌 , 董昭 , 等 . 基于图神经网络的数字全息图重建 [J ] . 光学学报 , 2025 , 45 （ 07 ）: 66 — 74 .

Hu Xianfei , Gui Jinbin , Dong Zhao , et al . Digital hologram reconstruction based on graph neural networks [J ] . Acta Optica Sinica , 2025 , 45 （ 07 ）: 66 — 74 .

赵长春 . 基于注意力机制的计算全息图生成算法研究 [D ] . 重庆理工大学 , 2024 .

Zhao Changchun . Research on computational hologram generation algorithm based on attention mechanism [D ] . Chongqing University Of Technology , 2024 .

孙康生 , 韩超 , 秦海峰 , 等 . 基于注意力卷积神经网络的高质量全息图快速生成算法 [J ] . 物理学报 2025 , 74 （ 08 ）: 169 — 178 .

Sun Kangsheng , Han Chao , Qin Haifeng , et al . A fast generation algorithm for high-quality holograms based on attention convolutional neural networks [J ] . Acta Physica Sinica , 2025 , 74 （ 08 ）: 169 — 178 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Wavefront correction method based on wavefront prediction model

Ultrasound multi-object tracking method based on multi-step matching and system construction

Intelligent fill level monitoring device for ball mill vibration signals based on fiber optic microphones

Nonlinear impairment compensation method for Long-haul Fiber-optic communication system

3D Confocal Raman Spectroscopic High-Resolution imaging method based on RDCAN deep learning model

Related Author

DU Guangxin

MA Yulei

AI Danni

ZHU Xinglong

SHI Xiaoju

XING Jichuan

ZHU Binzhi

WANG Daoping

Related Institution

Department of Computer and Information Engineering, Xinxiang University

Chinese People's Liberation Army Unit

Beijing Institute of Technology, School of Optics and Photonics

GuoNeng Sichuan Huaying Mountain Power Generation Co., Ltd.

School of Optics and Photonics, Beijing Institute of Technology

⁰