该文提供了一种应用空间分辨高光谱技术(spatially resolved hyperspectral imaging,SR-HSI)快速无损检测牛奶脂肪含量的方法。该文采用空间分辨高光谱成像系统在500~1 000 nm内采集不同脂肪含量牛奶样本的空间分辨高光谱图像,结合漫射近似光传输模型反演提取其吸收(μa)和约化散射系数(μ′s)光谱。分别基于μa和μ′s光谱数据,采用全波段、遗传算法(genetic algorithm,GA)和竞争性自适应加权算法(competitive adaptive reweighed sampling,CARS)的特征波长筛选方法,建立偏最小二乘回归(partial least squares regression,PLSR)和最小二乘支持向量机(support vector machine,SVM)模型。结果表明,μ′s与牛奶脂肪含量之间表现出良好的线性相关性,平均相关系数为0.995。基于μ′s光谱数据,选择GA波长筛选方法结合SVM建立的模型对牛奶脂肪含量预测效果最佳,预测集R2P为0.951 3,预测集均方根误差(root mean square error of prediction,RMSEP)为0.055 5;空间分辨高光谱技术能有效表征牛奶样本的光学特性,为牛奶脂肪含量的定量分析提供了有效手段。
This study presents a rapid, non-destructive method for detecting milk fat content using spatially resolved hyperspectral imaging (SR-HSI) technology.A spatially resolved hyperspectral imaging system was used to capture hyperspectral images of milk samples with varying fat contents within the spectral range of 500-1 000 nm.The absorption coefficient (μa) and reduced scattering coefficient (μ′s) spectra were extracted by inverting the diffuse approximation light transport model.To enhance model accuracy, three feature wavelength selection methods were applied:full-spectrum analysis, genetic algorithm (GA), and competitive adaptive reweighted sampling (CARS).These methods were used to develop partial least squares regression (PLSR) and support vector machine (SVM) models for predicting milk fat content.The results demonstrated a strong linear correlation between μ′s and milk fat content, with an average correlation coefficient of 0.995.The optimal performance for predicting milk fat content was achieved using the SVM model combined with the GA wavelength selection method based on μ′s spectral data, yielding a prediction set R2P of 0.951 3 and an RMSEP of 0.055 5.These findings suggest that SR-HSI can effectively characterize the optical properties of milk samples, and this method provides a reliable and efficient tool for the quantitative analysis of milk fat content, offering a promising approach for quality control in dairy production and processing.
[1] YU Y, FU R Q, JIN C J, et al.Regulation of milk fat synthesis:Key genes and microbial functions[J].Microorganisms, 2024, 12(11):2302.
[2] 曾云凤, 周振新, 刘昨, 等.盖勃法测定牛奶中脂肪含量的优化[J].中国乳业, 2024(7):98-101.
ZENG Y F, ZHOU Z X, LIU Z, et al.Optimization of gerber method for determination of fat content in milk[J].China Dairy, 2024(7):98-101.
[3] 杨彪. 生鲜和均质牛乳主要成分光谱检测方法及便携式仪器研发[D].咸阳:西北农林科技大学, 2021.
YANG B.Detection methods on raw and homogenized milk main compositions using spectroscopy and development of portable milk analyzer[D].Xianyang:Northwest A&F University, 2021.
[4] 李欣, 刘江平, 黄清, 等.改进的鲸鱼优化算法对牛奶脂肪含量预测模型的优化[J].光谱学与光谱分析, 2023, 43(9):2779-2784.
LI X, LIU J P, HUANG Q, et al.Optimization of prediction model for milk fat content based on improved whale optimization algorithm[J].Spectroscopy and Spectral Analysis, 2023, 43(9):2779-2784.
[5] JIN X, XIAO Z Y, XIAO D X, et al.Quantitative inversion model of protein and fat content in milk based on hyperspectral techniques[J].International Dairy Journal, 2022, 134:105467.
[6] GASTÉLUM-BARRIOS A, SOTO-ZARAZÚA G M, ESCAMILLA-GARCÍA A, et al.Optical methods based on ultraviolet, visible, and near-infrared spectra to estimate fat and protein in raw milk:A review[J].Sensors, 2020, 20(12):3356.
[7] MUHAMAD KAMIL N A I,WAN ISMAIL W Z, BALAKRISHNAN S R, et al.Study on optical properties of milk based on light propagation theory[J].Journal of Physics:Conference Series, 2021, 2071(1):012006.
[8] LUO Y F, DAI L Y, JIANG X, et al.Measurement of absorption and scattering properties of milk using a hyperspectral spatial frequency domain imaging system[J].Journal of Food Measurement and Characterization, 2022, 16(1):753-761.
[9] BOGOMOLOV A, BELIKOVA V, GALYANIN V, et al.Reference-free spectroscopic determination of fat and protein in milk in the visible and near infrared region below 1000 nm using spatially resolved diffuse reflectance fiber probe[J].Talanta, 2017, 167:563-572.
[10] LU Y Z, HUANG Y P, LU R F.Innovative hyperspectral imaging-based techniques for quality evaluation of fruits and vegetables:A review[J].Applied Sciences, 2017, 7(2):189.
[11] 张文祥, 潘嘹, 卢立新.包装生鲜牛肉pH值的高光谱无损检测方法[J].食品与发酵工业, 2023, 49(22):273-278.
ZHANG W X, PAN L, LU L X.Hyperspectral non-destructive testing method for pH value of packaged fresh beef[J].Food and Fermentation Industries, 2023, 49(22):273-278.
[12] 孙宏伟, 彭彦昆, 王凡.空间分辨散射光谱多参数信息融合方法的生鲜肉嫩度无损检测[J].光谱学与光谱分析, 2019, 39(11):3365.
SUN H W, PENG Y K, WANG F.Nondestructive detection of pork tenderness using spatially resolved hyperspectral imaging technique based on multivariable statistical analysis[J].Spectroscopy and Spectral Analysis, 2019, 39(11):3365.
[13] MÄTZLER C.MATLAB functions for mie scattering and absorption, version 2[J].IAP Research Reports, 2002, 8(1):1-25.
[14] 杨宇, 邢永春, 朱启兵.基于高光谱散射技术的粉体堆密度无损检测[J].激光与光电子学进展, 2018, 55(1):13003.
YANG Y, XING Y C, ZHU Q B.Non-destructive detection of bulk density of powder using hyperspectral scattering technique[J].Laser & Optoelectronics Progress, 2018, 55(1):13003.
[15] QIN J W, LU R F.Hyperspectral diffuse reflectance imaging for rapid, noncontact measurement of the optical properties of turbid materials[J].Applied Optics, 2006, 45(32):8366-8373.
[16] GRAAFF R, AARNOUDSE J G, ZIJP J R, et al.Reduced light-scattering properties for mixtures of spherical particles:A simple approximation derived from Mie calculations[J].Applied Optics, 1992, 31(10):1370-1376.
[17] WANG A C, WEI X H.A sequential method for estimating the optical properties of two-layer agro-products from spatially-resolved diffuse reflectance:Simulation[J].Artificial Intelligence in Agriculture, 2019, 3:69-78.
[18] MOHAMMADI S, GOWEN A, LUO J N, et al.Prediction of milk composition using multivariate chemometric modelling of infrared, Raman, and fluorescence spectroscopic data:A review[J].Food Control, 2024, 165:110658.
[19] BUOIO E, COLOMBO V, IGHINA E, et al.Rapid classification of milk using a cost-effective near infrared spectroscopy device and variable cluster-support vector machine (VC-SVM) hybrid models[J].Foods, 2024, 13(20):3279.
[20] DING G, JIN K, CHEN X Y, et al.Non-destructive prediction of ready-to-eat kiwifruit firmness based on Fourier transform near-infrared spectroscopy[J].Postharvest Biology and Technology, 2024, 212:112908.
[21] KAMIL, NOR’AINI, Z S, WAN ISMAIL, W Z, et al.Investigating the quality of milk using spectrometry technique and scattering theory[J].Engineering Technology & Applied Science Research, 2021, 11(3):7111-7117.
[22] ABILDGAARD O H A, KAMRAN F, DAHL A B, et al.Non-invasive assessment of dairy products using spatially resolved diffuse reflectance spectroscopy[J].Applied Spectroscopy, 2015, 69(9):1096-1105.
[23] KAMIL N A I M, ZAKARIA I H, ISMAIL W Z W, et al.Investigating light propagation in full and skimmed milk based on spectroscopy and monte carlo analysis[J].Journal of Engineering and Applied Science, 2020, 67(8):2265-2278.
