Simultaneous estimation of multiple order phase derivatives using deep learning method in digital holographic interferometry

Optics and Lasers in Engineering, Год журнала: 2024, Номер 184, С. 108583 - 108583

Опубликована: Сен. 13, 2024

Язык: Английский

---

Deep learning-based single-shot lateral shearing interferometry

Optics and Lasers in Engineering, Год журнала: 2025, Номер 191, С. 109010 - 109010

Опубликована: Апрель 14, 2025

Язык: Английский

---

Non-contact automated defect detection using a deep learning approach in diffraction phase microscopy

Optics Continuum, Год журнала: 2023, Номер 2(11), С. 2421 - 2421

Опубликована: Ноя. 7, 2023

Precision measurement of defects from optical fringe patterns is a problem significant practical relevance in non-destructive metrology. In this paper, we propose robust deep learning approach based on atrous convolution neural network model for defect detection noisy obtained diffraction phase microscopy. The utilizes the wrapped pattern as an input and generates binary image depicting non-defect regions output. effectiveness proposed validated through numerical simulations various under different noise levels. Furthermore, application technique identifying microscopy experiments also confirmed.

Язык: Английский

---

Research on the Identification of Bridge Structural Damage Using Variational Mode Decomposition and Convolutional Self-Attention Neural Networks

Applied Sciences, Год журнала: 2023, Номер 13(21), С. 12082 - 12082

Опубликована: Ноя. 6, 2023

Convolutional neural networks (CNN) are widely used for structural damage identification. However, the presence of environmental disturbances introduces noise into acquired acceleration response data, impairing performance CNN models. In this study, we apply empirical mode decomposition (EMD) and variational (VMD) to denoise data from a steel truss bridge. By comparing smoothness convergence obtained modal functions (IMFs) using EMD VMD, confirm effectiveness VMD in smoothing denoising bridge structure signals. Additionally, propose convolutional self-attention network (CSANN) model extract features identify denoised VMD. Comparative analysis CNN, LSTM, GRU models reveals that VMD-CSANN outperforms others terms localization identification accuracy. It also exhibits excellent when handling noise-contaminated with level 10%. These findings demonstrate efficacy proposed method identifying internal structures, while maintaining robustness during processing.

Язык: Английский

---

Wafer Edge Metrology and Inspection Technique Using Curved-Edge Diffractive Fringe Pattern Analysis

Journal of Manufacturing Science and Engineering, Год журнала: 2024, Номер 146(7)

Опубликована: Май 31, 2024

Abstract This paper introduces a novel wafer-edge quality inspection method based on analysis of curved-edge diffractive fringe patterns, which occur when light is incident and diffracts around the wafer edge. The proposed aims to identify various defect modes at edges, including particles, chipping, scratches, thin-film deposition, hybrid cases. diffraction patterns formed behind edge are influenced by factors, geometry, topography, presence defects. In this study, were obtained from two approaches: (1) single photodiode collected interferometric scanning (2) an imaging device coupled with objective lens captured image. first approach allowed apex characterization, while second enabled simultaneous localization characterization along bevels directions. analyzed both statistical feature extraction wavelet transform; corresponding features also evaluated through logarithm approximation. sum, methods can effectively characterize modes. Their potential lies in their applicability online metrology applications, thereby contributing advancement manufacturing processes.

Язык: Английский

---

Neural network based subspace analysis for estimation of phase derivatives from noisy interferograms

Опубликована: Янв. 1, 2024

This article introduces a robust phase derivative estimation method using deep learning-assisted subspace analysis. Simulation results validate the performance of proposed approach under severe noise conditions.

Язык: Английский

---

Non-destructive surface defect metrology using deep learning and diffraction phase microscopy

Опубликована: Янв. 1, 2024

We present an approach that utilizes a deep learning network to compute phase gradient for defect identification. The efficacy of this method is showcased through the analysis experimentally acquired noisy interferograms.

Язык: Английский

---

Deep learning-based automated defect detection in digital holographic microscopy

Опубликована: Янв. 1, 2024

The article introduces a defect identification method using digital holographic microscopy and deep learning. It utilizes wrapped phase from holograms to generate binary maps trained for high noise levels. Experimental results validate its efficacy.

Язык: Английский

---

Non-destructive fault diagnosis of electronic interconnects by learning signal patterns of reflection coefficient in the frequency domain

Microelectronics Reliability, Год журнала: 2024, Номер 162, С. 115518 - 115518

Опубликована: Окт. 14, 2024

Язык: Английский

---