Rapid identification of the green tea geographical origin and processing month based on near-infrared hyperspectral imaging combined with chemometrics

Ginkgo biloba Sex Identification Methods Using Hyperspectral Imaging and Machine Learning

Ginkgo biloba L. is a rare dioecious species that is valued for its diverse applications and is cultivated globally. This study aimed to develop a rapid and effective method for determining the sex of a Ginkgo biloba. Green and yellow leaves representing annual growth stages were scanned with a hyperspectral imager, and classification models for RGB images, spectral features, and a fusion of spectral and image features were established. Initially, a ResNet101 model classified the RGB dataset using the proportional scaling-background expansion preprocessing method, achieving an accuracy of 90.27%. Further, machine learning algorithms like support vector machine (SVM), linear discriminant analysis (LDA), and subspace discriminant analysis (SDA) were applied. Optimal results were achieved with SVM and SDA in the green leaf stage and LDA in the yellow leaf stage, with prediction accuracies of 87.35% and 98.85%, respectively. To fully utilize the optimal model, a two-stage Period-Predetermined (PP) method was proposed, and a fusion dataset was built using the spectral and image features. The overall accuracy for the prediction set was as high as 96.30%. This is the first study to establish a standard technique framework for Ginkgo sex classification using hyperspectral imaging, offering an efficient tool for industrial and ecological applications and the potential for classifying other dioecious plants.

Combination of near-infrared spectroscopy with Wasserstein generative adversarial networks for rapidly detecting raw material quality for formula products

Near-infrared spectroscopy (NIRS) has emerged as a key technique for rapid quality detection owing to its fast, non-destructive, and eco-friendly characteristics. However, its practical implementation within the formulation industry is challenging owing to insufficient data, which renders model fitting difficult. The complexity of acquiring spectra and spectral reference values results in limited spectral data, aggravating the problem of low generalization, which diminishes model performance. To address this problem, we introduce what we believe to be a novel approach combining NIRS with Wasserstein generative adversarial networks (WGANs). Specifically, spectral data are collected from representative samples of raw material provided by a formula enterprise. Then, the WGAN augments the database by generating synthetic data resembling the raw spectral data. Finally, we establish various prediction models using the PLSR, SVR, LightGBM, and XGBoost algorithms. Experimental results show the NIRS-WGAN method significantly improves the performance of prediction models, with R2 and RMSE of 0.949 and 1.415 for the chemical components of sugar, respectively, and 0.922 and 0.243 for nicotine. The proposed framework effectively enhances the predictive capabilities of various models, addressing the issue caused by limited training data in NIRS prediction tasks.

Rapid identification of the geographical origin of Baimudan tea using a Multi-AdaBoost model integrated with Raman Spectroscopy

The potential of Multi-AdaBoost in spectral analysis is substantial, particularly when combined with weak classifiers and trained to develop into a robust classifier. Given the variable quality of Baimudan tea sourced from diverse regions, the novel application of Raman spectroscopy in conjunction with the Multi-AdaBoost model to analyze the geographic origin of Baimudan tea was introduced. Initially, Raman spectra of Baimudan tea from four distinct origins in Fujian province were gathered, namely Fuan (FA), Fuding (FD), Zhenghe (ZH), and Songxi (SX). Decision Tree (DT) and Support Vector Machine (SVM) models were employed as fitting classifiers to construct the Multi-AdaBoost-DT and Multi-AdaBoost-SVM models. The results demonstrated that the Multi-AdaBoost-DT model exhibited significantly improved recognition rates for FA, FD, ZH, and SX origin compared to the DT model, with the average recognition rate increasing from 86.46% to 91.67%. In contrast, the recognition rates for FA and SX origin in the Multi-AdaBoost-SVM model remained unchanged, attributed to the model having reached a local optimum. The recognition rates of FD origin increased from 91.67% to 95.83%, a significant improvement, while those of ZH origin escalated from 83.33% to 87.50%. The average recognition rate increased from 92.71% to 94.79%. Additionally, Multi-AdaBoost-SVM and Multi-AdaBoost-DT enhanced the sensitivity and specificity of the discrimination outcomes. These results corroborated the effectiveness of the proposed Multi-AdaBoost-SVM model in identifying the geographical origin of Baimudan tea. Moreover, the Multi-AdaBoost model demonstrates potential in elevating the discrimination accuracy of weak classifiers, which bodes well for its application in food authentication and quality control.

Varietal Authenticity Assessment of QTMJ Tea Using Non-Targeted Metabolomics and Multi-Elemental Analysis with Chemometrics

In this paper, a combination of non-targeted metabolomics and multi-element analysis was used to investigate the impact of five different cultivars on the sensory quality of QTMJ tea and identify candidate markers for varietal authenticity assessment. With chemometric analysis, a total of 54 differential metabolites were screened, with the abundances significantly varied in the tea cultivars. By contrast, the QTMJ tea from the Yaoshan Xiulv (XL) monovariety presents a much better sensory quality as result of the relatively more abundant anthocyanin glycosides and the lower levels of 2'-o-methyladenosine, denudatine, kynurenic acid and L-pipecolic acid. In addition, multi-elemental analysis found 14 significantly differential elements among the cultivars (VIP > 1 and p < 0.05). The differences and correlations of metabolites and elemental signatures of QTMJ tea between five cultivars were discussed using a Pearson correlation analysis. Element characteristics can be used as the best discriminant index for different cultivars of QTMJT, with a predictive accuracy of 100%.

Tea authentication and determination of chemical constituents using digital image-based fingerprint signatures and chemometrics

Tea (Camellia sinensis) fraud has been frequently identified and involves tampering with the labelling of inferior products or without geographical origin certification and even mixing them with superior quality teas to mask an adulteration. Consequently, economic losses and health damage to consumers are observed. Thus, a Chemometrics-assisted Color Histogram-based Analytical System (CACHAS) was employed a simple, cost-effective, reliable, and green analytical tool to screen the quality of teas. Data-Driven Soft Independent Modeling of Class Analogy was used to authenticate their geographical origin and category simultaneously, recognizing correctly all Argentinean and Sri Lankan black teas and Argentinean green teas. For the determination of moisture, total polyphenols, and caffeine, Partial Least Squares obtained satisfactory predictive abilities, with values of root mean squared error of prediction (RMSEP) of 0.50, 0.788, and 0.25 mg kg^-1, r_pred of 0.81, 0.902, and 0.81, and relative error of prediction (REP) of 6.38, 9.031, and 14.58%., respectively. CACHAS proved to be a good alternative tool for environmentally-friendly non-destructive chemical analysis.

A Study on Origin Traceability of White Tea (White Peony) Based on Near-Infrared Spectroscopy and Machine Learning Algorithms

Identifying the geographical origins of white tea is of significance because the quality and price of white tea from different production areas vary largely from different growing environment and climatic conditions. In this study, we used near-infrared spectroscopy (NIRS) with white tea (n = 579) to produce models to discriminate these origins under different conditions. Continuous wavelet transform (CWT), min-max normalization (Minmax), multiplicative scattering correction (MSC) and standard normal variables (SNV) were used to preprocess the original spectra (OS). The approaches of principal component analysis (PCA), linear discriminant analysis (LDA) and successive projection algorithm (SPA) were used for features extraction. Subsequently, identification models of white tea from different provinces of China (DPC), different districts of Fujian Province (DDFP) and authenticity of Fuding white tea (AFWT) were established by K-nearest neighbors (KNN), random forest (RF) and support vector machine (SVM) algorithms. Among the established models, DPC-CWT-LDA-KNN, DDFP-OS-LDA-KNN and AFWT-OS-LDA-KNN have the best performances, with recognition accuracies of 88.97%, 93.88% and 97.96%, respectively; the area under curve (AUC) values were 0.85, 0.93 and 0.98, respectively. The research revealed that NIRS with machine learning algorithms can be an effective tool for the geographical origin traceability of white tea.

Discrimination of the harvesting season of green tea by alcohol/salt-based aqueous two-phase systems combined with chemometric analysis

The flavor and aroma quality of green tea are closely related to the harvest season. The aim of this study was to identify the harvesting season of green tea by alcohol/salt-based aqueous two-phase system (ATPS) combined with chemometric analysis. In this paper, the single factor experiments (SFM) and response surface methodology (RSM) optimization were designed to investigate and select the optimal ATPS. A total of 180 green tea samples were studied in this work, including 86 spring tea and 94 autumn tea. After the active components in green tea samples were extracted by the optimal ethanol/(NH₄)₂SO₄ ATPS, the qualitative and quantitative analysis was realized based on HPLC-DAD combined with alternating trilinear decomposition-assisted multivariate curve resolution (ATLD-MCR) algorithm, with satisfactory spiked recoveries (86.00 %-112.45 %). The quantitative results obtained from ATLD-MCR model were subjected to chemometric pattern recognition analysis. The constructed partial least squares-discriminant analysis (PLS-DA) and orthogonal partial least squares-discriminant analysis (OPLS-DA) models showed better results than the principal component analysis (PCA) model, and the R²X_cum values (>0.835) and R²Y_cum (>0.937) were close to 1, the Q²_cum values were greater than 0.75 (>0.933), and the differences between R²Y_cum and Q²_cum were not larger than 0.2, indicating excellent cross-validation prediction performance of the models. Furthermore, the classification results based on the hierarchical clustering analysis (HCA) were consistent with the PCA, PLS-DA and OPLS-DA results, establishing a good correlation between tea active components and the harvesting seasons of green tea. Overall, the combination of ATPS and chemometric methods is accurate, sensitive, fast and reliable for the qualitative and quantitative determination of tea active components, providing guidance for the quality control of green tea.

Cellulose-Based Light-Management Films with Improved Properties Directly Fabricated from Green Tea

Tea polyphenols are a phenolic bioactive compound extracted from tea leaves and have been widely used as additives to prepare functional materials used in packaging, adsorption and energy fields. Nevertheless, tea polyphenols should be extracted first from the leaves before use, leading to energy consumption and the waste of tea. Therefore, completely and directly utilizing the tea leaf to fabricate novel composite materials is more attractive and meaningful. Herein, semi-transparent green-tea-based all-biomass light-management films with improved strength, a tunable haze (60–80%) and UV-shielding properties (24.23% for UVA and 4.45% for UVB) were directly manufactured from green tea by adding high-degree polymerization wood pulps to form entanglement networks. Additionally, the green-tea-based composite films can be produced on a large scale by adding green tea solution units to the existing continuous production process of pure cellulose films. Thus, a facile and feasible approach was proposed to realize the valorization of green tea by preparing green-tea-based all-biomass light-management films that have great prospects in flexible devices and energy-efficient buildings.