Black tea is one of the most widely consumed infusions worldwide, whose quality changes during storage due to physicochemical and color changes. However, quantitative prediction of key physicochemical parameters, such as pH and water activity (Aw), during storage has been scarcely explored using hyperspectral imaging and machine learning. This study evaluated visible–near infrared (VIS–NIR) spectra acquired by hyperspectral imaging (HSI) from samples stored under different conditions to predict pH and Aw. A reproducible machine learning pipeline (second-order Savitzky–Golay derivative + standard normal variant (SNV), train–test baseline evaluation followed by stratified cross-validation) was implemented, and PLS, elastic net, random forest, support vector regression (SVR)-radial basis function (RBF), and XGBoost were compared. SVR-RBF showed the best performance (pH R2 ≈ 0.94, Aw R2 ≈ 0.82). Consensus-based spectral attribution identified relevant bands in the visible (390–550 nm), associated with polyphenolic chromophores, and in the short NIR (870–960 and 980–1000 nm), linked to O–H and C–H overtones. Leave-one-band-out (LOBO) validation revealed minimal variations (ΔR2 ≈ 0), confirming informational redundancy and robustness against band exclusion. In conclusion, VIS–NIR combined with machine learning is a potentially effective and portable tool for monitoring critical quality parameters in black tea, reducing the need for recalibration and facilitating the implementation of compact sensors in at-line and on-line applications.
Citation: André Rodríguez-León, Jimy Oblitas. Band mapping in hyperspectral images for predicting physicochemical parameters in black tea (Camellia sinensis)[J]. AIMS Agriculture and Food, 2026, 11(3): 565-583. doi: 10.3934/agrfood.2026029
Black tea is one of the most widely consumed infusions worldwide, whose quality changes during storage due to physicochemical and color changes. However, quantitative prediction of key physicochemical parameters, such as pH and water activity (Aw), during storage has been scarcely explored using hyperspectral imaging and machine learning. This study evaluated visible–near infrared (VIS–NIR) spectra acquired by hyperspectral imaging (HSI) from samples stored under different conditions to predict pH and Aw. A reproducible machine learning pipeline (second-order Savitzky–Golay derivative + standard normal variant (SNV), train–test baseline evaluation followed by stratified cross-validation) was implemented, and PLS, elastic net, random forest, support vector regression (SVR)-radial basis function (RBF), and XGBoost were compared. SVR-RBF showed the best performance (pH R2 ≈ 0.94, Aw R2 ≈ 0.82). Consensus-based spectral attribution identified relevant bands in the visible (390–550 nm), associated with polyphenolic chromophores, and in the short NIR (870–960 and 980–1000 nm), linked to O–H and C–H overtones. Leave-one-band-out (LOBO) validation revealed minimal variations (ΔR2 ≈ 0), confirming informational redundancy and robustness against band exclusion. In conclusion, VIS–NIR combined with machine learning is a potentially effective and portable tool for monitoring critical quality parameters in black tea, reducing the need for recalibration and facilitating the implementation of compact sensors in at-line and on-line applications.
| [1] |
Zhang Y, Liao B, Gouda M, et al. (2025) Visualization of moisture distribution in stacked tea leaves on process flow line using hyperspectral imaging. Foods 14: 1551. https://doi.org/10.3390/foods14091551 doi: 10.3390/foods14091551
|
| [2] |
Luo X, Sun C, He Y, et al. (2023) Cross-cultivar prediction of quality indicators of tea based on VIS-NIR hyperspectral imaging. Ind Crop Prod 202: 117009. https://doi.org/10.1016/j.indcrop.2023.117009 doi: 10.1016/j.indcrop.2023.117009
|
| [3] |
Li D, Chen Q, Ouyang Q, et al. (2025) Advances of VIS/NIRS and imaging techniques assisted by AI for tea processing. Crit Rev Food Sci 65: 7575–7593. https://doi.org/10.1080/10408398.2025.2474183 doi: 10.1080/10408398.2025.2474183
|
| [4] |
Feng L, Wu B, Zhu S, et al. (2021) Application of visible/infrared spectroscopy and hyperspectral Imaging with machine learning techniques for identifying food varieties and geographical origins. Front Nutr 8. https://doi.org/10.3389/fnut.2021.680357 doi: 10.3389/fnut.2021.680357
|
| [5] |
Ye W, Xu W, Yan T, et al. (2022) Application of near-infrared spectroscopy and hyperspectral imaging combined with machine learning algorithms for quality inspection of grape: A review. Foods 12: 132. https://doi.org/10.3390/foods12010132 doi: 10.3390/foods12010132
|
| [6] |
Gong J, Chen G, Deng Y, et al. (2024) Non-destructive detection of tea polyphenols in Fu Brick tea based on hyperspectral imaging and improved PKO-SVR method. Agriculture 14: 1701. https://doi.org/10.3390/agriculture14101701 doi: 10.3390/agriculture14101701
|
| [7] |
Li F, Shen J, Yang Q, et al. (2024) Monitoring quality changes in green tea during storage: A hyperspectral imaging method. Food Chem X 23: 101538. https://doi.org/10.1016/j.fochx.2024.101538 doi: 10.1016/j.fochx.2024.101538
|
| [8] |
Rodríguez-León A, Oblitas J, Quevedo-Olaya JL, et al. (2026) Non-destructive detection of Elasmopalpus lignosellus infestation in fresh asparagus using VIS–NIR hyperspectral imaging and machine learning. Foods 15: 355. https://doi.org/10.3390/foods15020355 doi: 10.3390/foods15020355
|
| [9] | Campbell GS, Galloway M, Campbell Z (2022) Measurement of water activity in the presence of high volatile concentrations using a tunable diode laser—single laboratory validation, first action 2021.04. J AOAC Int 105: 649–656. https://doi.org/10.1093/jaoacint/qsac003 |
| [10] |
Delgado M, Vanegas M, Delgado G (2007) Metrología química I: Calibración de un pHmetro y control de calidad. Universitas 1: 14–20. https://doi.org/10.5377/universitas.v1i1.1627 doi: 10.5377/universitas.v1i1.1627
|
| [11] |
Felizzato G, Bagnulo E, Tapparo G, et al. (2026) Methodological challenges in machine learning and deep learning applied to food analysis: A critical review. J Chromatogr A 1783: 467176. https://doi.org/10.1016/j.chroma.2026.467176 doi: 10.1016/j.chroma.2026.467176
|
| [12] |
Ban Y, Ren Y, Dong W, et al. (2026) Deep neural network frameworks (DNNFs) for enhanced non-destructive spectroscopic quantification in foods: From hybrid pipelines to end-to-end modelling. Trends Food Sci Tech 175: 105834. https://doi.org/10.1016/j.tifs.2026.105834 doi: 10.1016/j.tifs.2026.105834
|
| [13] |
Gillariose J, Joseph J, Chesneau C (2026) Lasso and Ridge regression: A comprehensive review of applications and developments in machine learning. Int J Data Sci Anal 21: 7. https://doi.org/10.1007/s41060-025-00957-y doi: 10.1007/s41060-025-00957-y
|
| [14] |
Zheng Q, Li W, Gao X (2021) The effect of storage time on tea polyphenols, catechin compounds, total flavones and the biological activity of Ya'an Tibetan tea (Camellia sinensis). J Food Process Pres 45. https://doi.org/10.1111/jfpp.16004 doi: 10.1111/jfpp.16004
|
| [15] |
Pujapanda R, Bashir O, Shams R, et al. (2025) Influence of carrier agents on nutritional, chemical, and antioxidant profiles of fruit juices during drying processes: A comprehensive review. DIFO 5: 228. https://doi.org/10.1007/s44187-025-00504-5 doi: 10.1007/s44187-025-00504-5
|
| [16] |
Saputra A, Herpandi H, Supriadi A, et al. (2025) Spectroscopic and chemometric characterization of storage-induced changes in pempek palembang using FT-IR and PCA. Int J Design Nat Ecodynamics 20: 1323–1332. https://doi.org/10.18280/ijdne.200612 doi: 10.18280/ijdne.200612
|
| [17] |
Giorgini E, Notarstefano V, Foligni R, et al. (2023) First ATR-FTIR characterization of black, green and white teas (Camellia sinensis) from european tea gardens: A PCA analysis to differentiate leaves from the In-Cup infusion. Foods 13: 109. https://doi.org/10.3390/foods13010109 doi: 10.3390/foods13010109
|
| [18] |
Falcioni R, Moriwaki T, Sversut M (2022) Classification and prediction by pigment content in lettuce (Lactuca sativa L.) varieties using machine learning and ATR-FTIR spectroscopy. Plants 11: 3413. https://doi.org/10.3390/plants11243413 doi: 10.3390/plants11243413
|
| [19] |
de Barros AT, Santos J, Silveira G, et al. (2025) Partial least squares methods in open-source software for identification of adulterated creatine using infrared spectroscopy. Food Chem 483: 143998. https://doi.org/10.1016/j.foodchem.2025.143998 doi: 10.1016/j.foodchem.2025.143998
|
| [20] |
Elamshity M, Alhamdan A (2025) Development and prediction of a non-destructive quality index (Qi) for stored date fruits using VIS–NIR spectroscopy and artificial neural networks. Foods 14: 3060. https://doi.org/10.3390/foods14173060 doi: 10.3390/foods14173060
|
| [21] | Liu J, Luan X, Liu F (2024) Full wavelength selection and deep spectra feature extraction network for near-infrared spectroscopy quantitative analysis, In: 2024 IEEE 13th Data Driven Control and Learning Systems Conference (DDCLS), IEEE, 1350–1355. https://doi.org/10.1109/DDCLS61622.2024.10606701 |
| [22] |
Malegori C, Muncan J, Mustorgi E, et al. (2022) Analysing the water spectral pattern by near-infrared spectroscopy and chemometrics as a dynamic multidimensional biomarker in preservation: rice germ storage monitoring. Spectrochim Acta A 265: 120396. https://doi.org/10.1016/j.saa.2021.120396 doi: 10.1016/j.saa.2021.120396
|
| [23] |
Visnupriyan R, Flanagan BM, Harper KJ, et al. (2024) Near infrared spectroscopy combined with chemometrics as tool to monitor starch hydrolysis. Carbohydr Polym 324: 121469. https://doi.org/10.1016/j.carbpol.2023.121469 doi: 10.1016/j.carbpol.2023.121469
|
| [24] |
Qi W, Jiang Q, Ma T, et al. (2025) Near-Infrared spectroscopy combined with chemometrics for liquor product quality assessment: A review. Foods 14: 2992. https://doi.org/10.3390/foods14172992 doi: 10.3390/foods14172992
|