EUVIMEDEUVIMED entdecken
Crossref

Single-shot fringe projection profilometry based on multi-task learning: efficient depth reconstruction without explicit system calibrations

Xiaopin Zhong, Junhao Huang, Yanhua Li, Jifeng Chen, Yibin Tian, Zongze Wu

Physica Scripta · 01.02.2025 · 100 (2) · 026005

DOI: 10.1088/1402-4896/ada20a ↗

Kurz gefasst

Fachartikel in der Zeitschrift Physica Scripta, online erschienen am 10.01.2025. Verfasst von Xiaopin Zhong und 5 weiteren Autor:innen. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 57 Literaturangaben. Gefördert von: Guangdong Major Project of Basic and Applied Basic Research, Equipments Development Project of the National Natural Science Foundation of China, The General Program of National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Fringe projection profilometry (FPP) typically captures multiple fringe patterns projected onto an object’s surface to reconstruct one frame of its 3D profile. Efficiency can be greatly improved by using only a single projected fringe, which is important for real-time applications. However, it is generally believed that there is insufficient information for reliable depth reconstruction from a single-shot fringe image. In this study, we propose a multi-task learning approach to improve the accuracy and robustness of depth reconstruction from single-shot FPP, eliminating the need for tedious explicit imaging system calibration. The proposed approach was extensively validated on both synthetic and real-world datasets, and compared with other state-of-the-art deep learning methods. Experimental results demonstrate that the proposed multi-task learning method for single-shot calibrationless FPP overcomes the limitations of traditional FPP and outperforms previous deep learning methods.

Förderung

  • Guangdong Major Project of Basic and Applied Basic Research
  • Equipments Development Project of the National Natural Science Foundation of China
  • The General Program of National Natural Science Foundation of China
  • Shenzhen Fundamental Research Fund
  • Equipment Development Project of Shenzhen University
  • Shenzhen University Program for Excellent Research

Zitieren

APA

Zhong, X., Huang, J., Li, Y., Chen, J., Tian, Y., & Wu, Z. (2025). Single-shot fringe projection profilometry based on multi-task learning: efficient depth reconstruction without explicit system calibrations. Physica Scripta, 100(2), 026005. https://doi.org/10.1088/1402-4896/ada20a

Vancouver

Zhong X, Huang J, Li Y, Chen J, Tian Y, Wu Z. Single-shot fringe projection profilometry based on multi-task learning: efficient depth reconstruction without explicit system calibrations. Physica Scripta. 2025;100(2):026005. doi:10.1088/1402-4896/ada20a

BibTeX
@article{Zhong2025,
  author = {Zhong, Xiaopin and Huang, Junhao and Li, Yanhua and Chen, Jifeng and Tian, Yibin and Wu, Zongze},
  title = {Single-shot fringe projection profilometry based on multi-task learning: efficient depth reconstruction without explicit system calibrations},
  journal = {Physica Scripta},
  year = {2025},
  volume = {100},
  number = {2},
  pages = {026005},
  doi = {10.1088/1402-4896/ada20a}
}
Kostenlos starten →