EUVIMEDEUVIMED entdecken
Crossref

Probabilistic network-based method for underwater vision measurement

Xu Wang, Mei Zhang, Ziwei Wang, Haoyu Wang, GuiHua Li

Optical Engineering · 12.02.2025 · 64 (02) · 1-12

DOI: 10.1117/1.oe.64.2.024105 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Optical Engineering, erschienen am 12.02.2025. Verfasst von Xu Wang und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Aiming to solve the problem where underwater digital image correlation cannot obtain real high-quality images and the inadequacy of the traditional point estimation probabilistic network that cannot cope well with the water degradation of underwater pictures, we propose a probabilistic network-based method for underwater vision measurement. First, a conditional variational self-encoder is combined with adaptive instance normalization to construct the augmented distribution. Then, an estimation process combining maximum probability estimation and Monte Carlo estimation is used to predict outcomes. The bias can be reduced to some extent by learning to augment the distribution. Finally, the captured images of the underwater calibration plates are passed into the model for processing to obtain images with enhanced color, improved contrast, improved sharpness, and less distortion in the image. Furthermore, the feasibility and reliability of the method are verified by underwater camera calibration experiments, comparison experiments, and underwater measurement experiments. Results show that the underwater reprojection error obtained by the proposed method is small, which is better than that of the data-driven implicit calibration method for underwater stereo cameras and the genetic algorithm-based parameter optimization method, and it works well for underwater measurement with high measurement accuracy.

Zitieren

APA

Wang, X., Zhang, M., Wang, Z., Wang, H., & Li, G. (2025). Probabilistic network-based method for underwater vision measurement. Optical Engineering, 64(02), 1-12. https://doi.org/10.1117/1.oe.64.2.024105

Vancouver

Wang X, Zhang M, Wang Z, Wang H, Li G. Probabilistic network-based method for underwater vision measurement. Optical Engineering. 2025;64(02):1-12. doi:10.1117/1.oe.64.2.024105

BibTeX
@article{Wang2025,
  author = {Wang, Xu and Zhang, Mei and Wang, Ziwei and Wang, Haoyu and Li, GuiHua},
  title = {Probabilistic network-based method for underwater vision measurement},
  journal = {Optical Engineering},
  year = {2025},
  volume = {64},
  number = {02},
  pages = {1--12},
  doi = {10.1117/1.oe.64.2.024105}
}
Kostenlos starten →