Robust phase unwrapping algorithm based on Zernike polynomial fitting and Swin-Transformer network

Zhao, Zixin; Zhou, Menghang; Du, Yijun; Li, Junxiang; Fan, Chen; Zhang, Xuchao; Xiang, Wei; Zhao, Hong

doi:10.1088/1361-6501/ac4ac2

Cited by 6 publications

(2 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this study, DCT unwrapping [20][21][22] is used to unwrap a wrapped phase. After phase gradient smoothing correction, DCT phase unwrapping is performed using the corrected phase gradient value.…”

Section: Principle Of Dct Unwrappingmentioning

confidence: 99%

Discrete cosine transformation unwrapping algorithm based on phase gradient smoothing correction

et al. 2023

View full text Add to dashboard Cite

At present, in the shear speckle interference technique for vehicle tire defect detection, the key is to accurately reflect the real phase of object motion when there are interference factors. Thus, this study proposes a discrete cosine transformation (DCT) unwrapping algorithm based on phase gradient smoothing correction, and analyzes the feasibility of the proposed method through several experiments and numerical simulation calculations. The proposed method replaces abnormal phase gradients caused by noise with calculated weighted moving average phase gradients, bringing phases closer to real values. Finally, the DCT unwrapping method is used to obtain the final solution of phase. A phase recovered using the proposed method is closer to the real value than that recovered using the traditional DCT unwrapping algorithm. Compared to the traditional DCT unwrapping algorithm, the root mean square error accuracy of the proposed method is improved by 10 to 100 times. And after many experiments to record the average running speed, it can be obtained that the time-consumption of the two algorithms only differs by ∼0.1 to 0.3 s.

show abstract

Section: Principle Of Dct Unwrappingmentioning

confidence: 99%

Discrete cosine transformation unwrapping algorithm based on phase gradient smoothing correction

et al. 2023

View full text Add to dashboard Cite

show abstract

“…Sun et al [27] proposed another architecture using a channel wise cross fusion transformer with UNET for phase unwrapping in single shot fringe projection profilometry. Zhao et al [28] have combined Swin Transformer with Zernike polynomial fitting for phase unwrapping. Here Zernike polynomial coefficients are predicted by the Swin Transformer network.…”

Section: Introductionmentioning

confidence: 99%

Transformer based deep learning hybrid architecture for phase unwrapping

Bujagouni,

Pradhan

2024

Phys. Scr.

View full text Add to dashboard Cite

A deep learning Hybrid architecture for phase unwrapping has been proposed. The hybrid architecture is based on integration of Convolutional Neural Networks (CNN) with Vision Transformer. The performance of Hybrid architecture/network in phase unwrapping is compared against CNN based standard UNET network. Structural Similarity Index (SSIM) and Root Mean Square Error (RMSE) have been used as performance metrics to assess the performance of these deep learning networks for phase unwrapping. To train and test the networks, dataset with high mean Entropy has been generated using Gaussian filtering of random noise in Fourier plane. The Hybrid architecture is tested on test dataset and is found to have superior performance metrics against the UNET network. Their performance is also tested in noisy environment with various noise levels and Hybrid architecture demonstrated better anti-noise capability than UNET network. Hybrid architecture was successfully validated in real world scenario using experimental data from custom built Digital Holographic Microscope. With the advent of newer architectures and hardware, Deep learning networks can further improve the performance in solving inverse problems.

show abstract

Applying machine learning to optical metrology: a review

Xue,

Hooshmand,

Isa

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

This literature review investigates the integration of machine learning (ML) into optical metrology, unveiling enhancements in both efficiency and effectiveness of measurement processes. With a focus on phase demodulation, unwrapping, and phase-to-height conversion, the review highlights how ML algorithms have transformed traditional optical metrology techniques, offering improved speed, accuracy, and data processing capabilities. Efficiency improvements are underscored by advancements in data generation, intelligent sampling, and processing strategies, where ML algorithms have accelerated the metrological evaluations. Effectiveness is enhanced in measurement precision, with ML providing robust solutions to complex pattern recognition and noise reduction challenges. Additionally, the role of parallel computing using graphics processing units and field programmable gate arrays is emphasised, showcasing their importance in supporting the computationally intensive ML algorithms for real-time processing. This review culminates in identifying future research directions, emphasising the potential of advanced ML models and broader applications within optical metrology. Through this investigation, the review articulates a future where optical metrology, empowered by ML, achieves improved levels of operational efficiency and effectiveness.

show abstract

Robust phase unwrapping algorithm based on Zernike polynomial fitting and Swin-Transformer network

Cited by 6 publications

References 33 publications

Discrete cosine transformation unwrapping algorithm based on phase gradient smoothing correction

Discrete cosine transformation unwrapping algorithm based on phase gradient smoothing correction

Transformer based deep learning hybrid architecture for phase unwrapping

Applying machine learning to optical metrology: a review

Contact Info

Product

Resources

About