EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

A Digital Image Correlation Method Based on Deep Learning under Low-Light Conditions

Xiao-Yong Liu, Jia-Ming Hu, Rong-Li Li, Guo-Qing Han, Song Gao, ZhaoPeng Hao, Yuan-Ze Huang, Dongwei Gu, Cheng Zhang

Measurement Science Review · 01.12.2026 · 26 (6) · 296-306

DOI: 10.2478/msr-2026-0036 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Measurement Science Review, online erschienen am 21.11.2026. Verfasst von Xiao-Yong Liu und 8 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Digital image correlation (DIC) is a widely used technique for deformation measurement, leveraging the analysis of grayscale information from surface speckle patterns to provide non-contact, full-field, and high-accuracy measurements. However, under low-light conditions, speckle images often suffer from insufficient brightness, low contrast, and significant noise, leading to degraded pattern quality and loss of critical texture information, which in turn severely undermines measurement accuracy. To address this limitation, this paper integrates deep learning-based low-light image enhancement with DIC, proposing an enhanced low-light DIC algorithm founded on EnlightenGAN, termed displacement measurement algorithm (EnDIC). A systematic performance comparison is conducted among four deep learning enhancement methods, including EnlightenGAN and Zero-Reference Deep Curve Estimation (Zero-DCE). Experimental results demonstrate that EnDIC outperforms the other methods in both image quality enhancement and displacement calculation accuracy. Furthermore, across simulated and real-world experiments under various low-light conditions, EnDIC exhibits higher measurement precision and superior robustness against illumination variations. The proposed approach thus offers a reliable and effective technical solution for engineering applications such as structural health monitoring in challenging lighting environments.

Zitieren

APA

Liu, X., Hu, J., Li, R., Han, G., Gao, S., Hao, Z., Huang, Y., Gu, D., & Zhang, C. (2026). A Digital Image Correlation Method Based on Deep Learning under Low-Light Conditions. Measurement Science Review, 26(6), 296-306. https://doi.org/10.2478/msr-2026-0036

Vancouver

Liu X, Hu J, Li R, Han G, Gao S, Hao Z, et al. A Digital Image Correlation Method Based on Deep Learning under Low-Light Conditions. Measurement Science Review. 2026;26(6):296-306. doi:10.2478/msr-2026-0036

BibTeX
@article{Liu2026,
  author = {Liu, Xiao-Yong and Hu, Jia-Ming and Li, Rong-Li and Han, Guo-Qing and Gao, Song and Hao, ZhaoPeng and Huang, Yuan-Ze and Gu, Dongwei and Zhang, Cheng},
  title = {A Digital Image Correlation Method Based on Deep Learning under Low-Light Conditions},
  journal = {Measurement Science Review},
  year = {2026},
  volume = {26},
  number = {6},
  pages = {296--306},
  doi = {10.2478/msr-2026-0036}
}
Kostenlos starten →