EUVIMEDEUVIMED entdecken
Crossref

Evaluation of spectral preprocessing strategies for LIBS-based classification of Fritillaria species

Chundong Xu, Yinghui Qi, Kai Wei, Geer Teng, Keliu Long, Rui Liu, Songzheng Zhang

Optics Continuum · 15.08.2026 · 5 (8) · 2514

DOI: 10.1364/optcon.604468 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Optics Continuum, online erschienen am 24.07.2026. Verfasst von Chundong Xu und 6 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Gefördert von: Jiangxi University of Science and Technology, Natural Science Foundation of Jiangxi Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate identification of different types of Fritillaria is of great significance for quality control and clinical application of traditional Chinese medicine. This study systematically analyzed the classification performance of three Fritillaria samples using laser-induced breakdown spectroscopy (LIBS) combined with different data preprocessing methods and compared three strategies: standard normal variable (SNV), area normalization, and maximum value normalization. Firstly, by comparing the original spectra with the relative standard deviations (RSDs) after preprocessing, it was found that the RSDs decreased, indicating that the preprocessing effectively suppressed spectral fluctuations. Further analysis, combined with principal component analysis (PCA) and clustering evaluation indicators, including the Calinski-Harabasz (CH) index and Davies-Bouldin index (DBI), revealed that different preprocessing methods had different effects on the structure of the feature space. Among them, SNV showed a relatively more balanced performance between intra-class consistency and inter-class separation. Finally, decision tree (DT), support vector machine (SVM), and random forest (RF) models were used to validate different preprocessing methods. The results showed that all preprocessing methods improved the classification accuracy to varying degrees. Among them, the combination of SNV and SVM improved the most significantly, with the correct classification rate increasing from 90.89% to 94.56% (p = 2.31 × 10 −6 ). In conclusion, reasonable spectral preprocessing can significantly improve the classification performance of Fritillaria species based on LIBS, providing a reference for the selection of preprocessing strategies in the spectral analysis of traditional Chinese medicine.

Förderung

  • Jiangxi University of Science and Technology
  • Natural Science Foundation of Jiangxi Province

Zitieren

APA

Xu, C., Qi, Y., Wei, K., Teng, G., Long, K., Liu, R., & Zhang, S. (2026). Evaluation of spectral preprocessing strategies for LIBS-based classification of Fritillaria species. Optics Continuum, 5(8), 2514. https://doi.org/10.1364/optcon.604468

Vancouver

Xu C, Qi Y, Wei K, Teng G, Long K, Liu R, et al. Evaluation of spectral preprocessing strategies for LIBS-based classification of Fritillaria species. Optics Continuum. 2026;5(8):2514. doi:10.1364/optcon.604468

BibTeX
@article{Xu2026,
  author = {Xu, Chundong and Qi, Yinghui and Wei, Kai and Teng, Geer and Long, Keliu and Liu, Rui and Zhang, Songzheng},
  title = {Evaluation of spectral preprocessing strategies for LIBS-based classification of Fritillaria species},
  journal = {Optics Continuum},
  year = {2026},
  volume = {5},
  number = {8},
  pages = {2514},
  doi = {10.1364/optcon.604468}
}
Kostenlos starten →