Differentiable Imaging: A New Tool for Computational Optical Imaging

Ni, Chen; Cao, Liangcai; Poon, Ting‐Chung; Lee, Byoungho; Lam, Edmund Y.

doi:10.1002/apxr.202200118

Cited by 18 publications

(5 citation statements)

References 95 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The effectiveness of this method has been confirmed through experimental results. Given the prevalence of misalignment issues in computational imaging [9,10], we anticipate that our approach could serve as a valuable reference for other computational imaging methodologies. Funding: National Science Foundation of China (62275178), Chengdu Science and Technology Program (2022-GH02-00016-HZ).…”

Section: Discussionmentioning

confidence: 99%

Adaptive cascade calibrated multi-plane phase retrieval

Wang,

Chen

2024

Preprint

View full text Add to dashboard Cite

show abstract

Section: Discussionmentioning

confidence: 99%

Adaptive cascade calibrated multi-plane phase retrieval

Wang,

Chen

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…As a stark contrast, digital inline holography (DIH) has emerged with immense significance in modern scientific imaging systems with applications ranging from biology and medicine to materials science and engineering. [1][2][3][4][5] By eliminating the need for physical lenses, DIH simplifies the imaging setup, alleviates the painstaking alignment efforts and circumvents optical aberrations that are inevitable in conventional lens-based imaging systems. These advantages make DIH particularly promising for portable and field research applications, which opens up new possibilities for scientific exploration in previously inaccessible environments and real-time monitoring such as micro-plastic investigation in marine system.…”

Section: Introductionmentioning

confidence: 99%

Single-shot digital holography with improved twin-image noise suppression using a diffusion-based generative model

Zhang,

Liu,

Lam

2024

Computational Optical Imaging and Artificial Intelligence in Biomedical Sciences

View full text Add to dashboard Cite

Due to the loss of phase information in images captured by intensity-only measurements, the numerical reconstruction of inline digital holographic imaging suffers from the undesirable twin-image artifact. This artifact presents as an out-of-focus conjugate at the virtual image plane and reduces the reconstruction quality. In this work, we propose a diffusion-based generative model that eliminates such defocus noise in single-shot inline digital holography. The diffusion-based generative model learns the implicit prior of the underlying data distribution by progressively injecting random noise in data and then generating high-quality samples by reversing this process. Although the diffusion model has been successful in various challenging tasks in computer vision, its potential in scientific imaging has not been fully explored yet, and one challenge is the inherent randomness in its reverse sampling process. To address this issue, we incorporate the underlying physics of image formation as a prior, which constrains the possible samples from the data distribution. Specifically, we include an extra gradient correction step in each reverse sampling process to introduce data consistency and generate better results. We demonstrate the feasibility of our approach using simulated and experimental holograms and compare our results with previous methods. Our model recovers detailed object information and significantly suppresses the twin-image noise. The proposed method is explainable, generalizable, and transferable to other samples from various distributions, making it a promising tool for digital holographic reconstruction.

show abstract

“…Therefore, accounting for the model mismatch is crucial to improve the reliability and applicability of CI. 7…”

Section: Introductionmentioning

confidence: 99%

Toward robust super-resolution imaging: A low-rank approximation approach for pattern-illuminated Fourier ptychography

Zhang,

Wei,

Yang

et al. 2024

APL Photonics

View full text Add to dashboard Cite

Pattern-illuminated Fourier ptychography (piFP) is an elegant combination of structured illumination imaging and a Fourier ptychographic algorithm with the ability to image beyond the diffraction limit of the employed optics. Artifact-free piFP super-resolution reconstruction requires a high level of stability in the illumination pattern. However, unpredictable pattern variation occurs in the presence of environment perturbation, intensity fluctuation, and pointing instability at the source, leading to declines in image reconstruction quality. To address this issue, we present an efficient and robust piFP algorithm based on low-rank approximation (LRA-piFP), which relaxes the requirement for the stability of illumination patterns. This LRA-piFP method can model frame-wise pattern variation during a full scan, thus improve the reconstruction quality significantly. We take numerical simulations and proof-of-principle experiments with both long-range imaging and microscopy for demonstrations. Results show that the LRA-piFP method can handle different kinds of pattern variation and outperforms other state-of-the-art techniques in terms of reconstruction quality and resolution improvement. Our method provides effective experimental robustness to piFP with a natural algorithmic extension, paving the way for its application in both macroscopic and microscopic imaging.

show abstract

Differentiable Imaging: A New Tool for Computational Optical Imaging

Cited by 18 publications

References 95 publications

Adaptive cascade calibrated multi-plane phase retrieval

Adaptive cascade calibrated multi-plane phase retrieval

Single-shot digital holography with improved twin-image noise suppression using a diffusion-based generative model

Toward robust super-resolution imaging: A low-rank approximation approach for pattern-illuminated Fourier ptychography

Contact Info

Product

Resources

About