.png){kind=icon}
.png){kind=icon}
Return
Fast and Efficient Phase Unwrapping Method Based on Deep Learning
.png){kind=icon}
OA
AI
DOI：10.3788/LOP250680.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In recent years, the structured light three-dimensional (3D) measurement technology has been widely used due to its non-contact, fast, and efficient features. As a key component in structured light 3D measurement technology, phase unwrapping has a crucial impact on the measurement results. However, the current phase unwrapping method based on deep learning have problems such as large parameter numbers and high computational complexity. To solve these problems, this study proposes a fast and efficient phase unwrapping method. A combination of 3x1 and 1x3 convolutions is used to replace the standard 3x3 convolution to reduce the number of parameters in the model. To address the problem of high computational complexity, a group multi-axis Hadamard product attention module is introduced, and a 1D group multi-axis Hadamard product attention module is designed. The performance of the proposed method and several typical deep learning-based phase unwrapping methods is evaluated on experimental data. The experimental results show that the parameter of the constructed model is approximately 0.24x10(6), and the mean absolute error and root mean square error of the proposed method are 0.3309 and 0.8319, respectively. The proposed method achieves good precision with less parameter numbers, verifying the feasibility and effectiveness.
Keywords:
fringe projection profilometry
three-dimensional morphology
phase unwrapping
deep learning
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
L
IF:
1
Papers:
91
Citations:
3.9K