View in EDS

Deep learning enabled single-shot 3D measurement for colorful object in fringe projection profilometry.

Saved in:
Bibliographic Details
Title: Deep learning enabled single-shot 3D measurement for colorful object in fringe projection profilometry.
Authors: Lin, Junyi1 (AUTHOR) ljy2004@hqu.edu.cn, Li, Yupeng (AUTHOR), Dou, Qi (AUTHOR), Huang, Changbiao (AUTHOR), Liu, Hua (AUTHOR), Lu, Ping (AUTHOR)
Source: Optics & Laser Technology. Jun2026, Vol. 198, pN.PAG-N.PAG. 1p.
Subjects: Shape measurement, Color, Deep learning, Artificial neural networks, Phase unwrapping (Digital image processing)
Abstract: The composite fringe projection profilometry based on deep learning has successfully achieved high efficiency and accuracy in single-shot 3D reconstruction. However, high-quality 3D measurement for colorful objects remains a challenging task in this field. Wrapped phase jumps and phase period ambiguity caused by color are the main reasons for the challenges. To solve these problems, this paper proposes an anti-color interference single-shot 3D reconstruction method (AIS3DRM) based on Haar-like composite fringe projection (HCFP) and dual-context scale-aware network (DCSA-Net), which utilizes HCFP to better reduce the phase jump and effectively avoid the phase period ambiguity in the colorful object measurement domain, and DCSA-Net to restore the non-ideal fringe patterns caused by color mutation. The deep-learning-enabled HCFP profilometry (DHCFPP) method is adopted to implement the three-frequency four-step phase-shifting method (TFPM) efficiently and accurately. DCSA-Net is an architecture based on U-Net, which integrates Spatial Frequency Fusion Attention Blocks (SFFABs) and an Adversarial Training Framework (ATF). SFFABs dynamically weight features through parallel branches to ensure robust performance under different object color conditions. ATF further optimizes the loss function by comparing the up-sampled output with real data. Our experimental results demonstrate that the proposed method outperforms existing methods in terms of the quality of the generated phase and the integrity of the point cloud for colorful object measurement. [ABSTRACT FROM AUTHOR]
Copyright of Optics & Laser Technology is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:The composite fringe projection profilometry based on deep learning has successfully achieved high efficiency and accuracy in single-shot 3D reconstruction. However, high-quality 3D measurement for colorful objects remains a challenging task in this field. Wrapped phase jumps and phase period ambiguity caused by color are the main reasons for the challenges. To solve these problems, this paper proposes an anti-color interference single-shot 3D reconstruction method (AIS3DRM) based on Haar-like composite fringe projection (HCFP) and dual-context scale-aware network (DCSA-Net), which utilizes HCFP to better reduce the phase jump and effectively avoid the phase period ambiguity in the colorful object measurement domain, and DCSA-Net to restore the non-ideal fringe patterns caused by color mutation. The deep-learning-enabled HCFP profilometry (DHCFPP) method is adopted to implement the three-frequency four-step phase-shifting method (TFPM) efficiently and accurately. DCSA-Net is an architecture based on U-Net, which integrates Spatial Frequency Fusion Attention Blocks (SFFABs) and an Adversarial Training Framework (ATF). SFFABs dynamically weight features through parallel branches to ensure robust performance under different object color conditions. ATF further optimizes the loss function by comparing the up-sampled output with real data. Our experimental results demonstrate that the proposed method outperforms existing methods in terms of the quality of the generated phase and the integrity of the point cloud for colorful object measurement. [ABSTRACT FROM AUTHOR]
ISSN:00303992
DOI:10.1016/j.optlastec.2026.114895