摘要(英) |
This study aims to explore the design of an optical system based on holography theory, utilizing a coaxial configuration to simplify equipment requirements and achieve miniaturization. Traditional holography requires extensive and complex optical equipment. In contrast, our designed system only needs a near-coaxial light source and sensor to capture an image area of nearly 30mm² in a single shot, equivalent to eight times the area under a 20x objective lens of a conventional microscope. In terms of image reconstruction, we innovatively proposed an algorithm that filters critical information through image masking based on the traditional holography reconstruction method (ASP), successfully solving the concentric circle issue in coaxial holography image reconstruction. Additionally, this study integrates the latest artificial intelligence technology by employing deep learning models to replace the mathematical theory of holography for image reconstruction. Through tests with different samples, the feasibility and effectiveness of this model have been demonstrated. We also combined noise reduction and resolution enhancement models, further improving the quality of reconstructed images. The comprehensive application of these technologies allows us to significantly enhance the clarity and accuracy of the images.
The results of this study demonstrate the great potential of combining lensless holographic microscopy systems with artificial intelligence technology, especially its application value in simplifying optical equipment, improving image quality, and achieving system miniaturization. In the future, we look forward to further optimizing the system and adding more advanced functions and applications to make it play a more significant role in various fields. Keywords: holography, deep learning, image reconstruction, image quality enhancement, miniaturized system. |
參考文獻 |
[1] Zhu, Hongying, et al. Optical imaging techniques for point-of-care diagnostics. Lab on a Chip, Vol.13, 2013, p. 51-67.
[2] McCarthy, J, et al. A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence. AI magazine, Vol.27,2006
[3] Flowers, Johnathon Charles. Strong and Weak AI: Deweyan Conssideration.AAAI Spring Symposium: Towards Conscious AI Systems.2019
[4] Ning, Cai, et al. A Review of Artificial Intelligence (AI) in Education from 2010 to 2020.ACM digital library Vol.01,2021
[5] Goodman, J.W., Introduction to Fourier optics. 2005: Roberts and Company Publishers.
[6] Seo, S, et al. Lensfree holographic imaging for on-chip cytometry and diagnostics. Lab on a Chip, Vol. 9, 2009, p. 777-87.
[7] Garcia-Sucerquia, J, et al, Digital in-line holographic microscopy. Applied optics, Vol. 45, 2006, p. 14.
[8] Yepes, I.S.V. and M.R.R. Gesualdi, Dynamic Digital Holography for recording and reconstruction of 3D images using optoelectronic devices. Journal of Microwaves. Optoelectronics and Electromagnetic Applications. Vol. 16, 2017, p. 801-815.
[9] J. R. Fienup. Phase retrieval algorithms: a comparison.
[10] L. J. Allen, et al. Phase retrieval from series of images obtained by defocus variation. Optics Communications,Vol.13,2001
[11] IEEE. A view of Artificial Neural Network. IEEE Xplore,2015
[12] Keiron O′Shea, Ryan Nash. An Introduction to Convolutional Neural Networks. arXiv:1511.08458,2015
[13] Xie, Yaochen, et al. Self-Supervised Learning of Graph Neural Networks: A Unified Review. IEEE Xplore 2023 p.2412-2429.
[14] M. J. Huang,Zi-Neng, He. Phase unwrapping through region-referenced algorithmand window-patching method. Optics Communications 2002
[15] Loic Denis, et al. Twin-image noise reduction by phase retrieval in in-linedigital holography. HAL ujm-00116996,2005
[16] Seo, S, et al,′Multi-color LUCAS: Lensfree On-chip Cytometry Using Tunable Monochromatic Illumination and Digital Noise Reduction. Cellular and Molecular Bioengineering,Vol. 1, 2008, p. 146-156.
[17] N. M. Kwok, et al. Gray world based color correction and intensity preservation for image enhancement. IEEE, Vol.2,2011
[18] Zhuoshi, Li. High Bandwidth-Utilization Digital Holographic Reconstruction Using an Untrained Neural Network. Applied sciences 10656,2022
[19] Zhiming Tian, et al. Lensless computational imaging with a hybridframework of holographic propagation and deeplearning. Optics Letter, Vol 47,No17,2022
[20] Luzhe Huang, etal. Self-supervised learning of hologramreconstruction using physics consistency. Nature machine intelligence.10.1038,2023
[21] Hanlong Chen, et al. Fourier Imager Network (FIN): A deep neuralnetwork for hologram reconstruction with superiorexternal generalization. Science and Application (2022)11:254
[22] D. A. Rymov, et al. Neural-network-enabled holographic imagereconstruction via amplitude and phaseextraction. Journal of Optical Technology. Vol.89, NO.9,2022
[23] Mikolaj Rogalski, et al. Physics-driven universal twin-image removal network for digital in line holographic microscopy. Optica 2024
[24] Christian Ledig, et al. Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network. arXir:1609.04802,2017
[25] Chao Dong, etal. Image Super-Resolution Using Deep Convolutional Networks. arXir:1501.00092,2015
[26] Wei-Sheng Lai, et al. Deep Laplacian Pyramid Networks for Fast and Accurate Super-Resolution. IEEE Xplore,2017
[27] Wang, Zhou, et al. Image quality assessment: from error visibility to structural similarity. IEEE Xplore :13.4,2004,p.600-612 |