Vis/NIR spectroscopy and machine learning model for counterfeit Citri Reticulatae Pericarpium identification

• Published in: Journal of food composition and analysis Vol. 148; p. 108240
• Main Authors: Zhang, Mingkun, Ma, Chao, Ma, Jianwei, Yuan, Yunxia, Huang, Jiayu, Yan, Yongyi
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.12.2025
• Subjects: Citri Reticulatae Pericarpium; Counterfeit identification; Machine learning; Spectral; Vis-NIR; Spectral; Counterfeit identification; Vis-NIR; Citri Reticulatae Pericarpium; Machine learning
• ISSN: 0889-1575
• DOI: 10.1016/j.jfca.2025.108240

Citri Reticulatae Pericarpium(CRP), due to its market demand and high price, is commonly subject to counterfeiting, with various methods of counterfeiting. In this study, visible/near-infrared (Vis/NIR) spectral images of authentic and counterfeited CRP were collected using a Vis/NIR spectrometer. Spectral data processing and machine learning classification models were utilized for classification. Preprocessing, dimensionality reduction, feature wavelength extraction, and machine learning were applied to classify CRP spectral data to address this issue. We apply the Seagull Optimization Algorithm to optimize SVM parameters, thereby proposing the SOASVM model. The results demonstrated that the proposed model could effectively and accurately distinguish between authentic and counterfeited CRP, as well as different methods of counterfeiting. Linear discriminant analysis(LDA) after data processing achieved the best performance, with the classification accuracy of up to 99.3% in test set when combined with the SOASVM model via cross-validation. This study provides optimized models for CRP counterfeiting classification, offering a non-destructive, precise, and effective method for distinguishing authentic from counterfeited CRP. [Display omitted] •The Spectra model non-destructively detects Citri Reticulatae Pericarpium counterfeits.•MSC and LDA are the optimal preprocessing and dimensionality reduction methods.•MSC-LDA-SOASVM combined with Vis/NIR achieves effective accuracy detection.