食品工业科技2025，Vol.46Issue(11)：302-312,11.DOI:10.13386/j.issn1002-0306.2024080030
基于高光谱成像技术和近红外光谱技术的金冠苹果货架期判别及其品质分析
Shelf Life Identification and Quality Analysis of Golden Delicious Apples Based on Hyperspectral Imaging and Near Infrared Spectroscopy
摘要
Abstract
In order to achieve non-destructive analysis of shelf life,soluble solid content(SSC)and pH of Golden Delicious apples,the spectral information of six different shelf life(postharvest 0,7,14,21,28 and 35 d)of apple was collected by hyperspectral imaging system(400～1000 nm)and near-infrared spectroscopy(800～2500 nm),respectively.The spectroscopy data was pro-processed by savitzky-golay(SGS),savitzky-golay first derivative(1D),standard normal variate(SNV),and area normalize(Normalize),competitive adaptive reweighted sampling aglorithm(CARS)and uninformative variable elimination(UVE)were used to extract characteristic wavelengths,and the shelf-life classification models were established by back propagation neural network(BP)and least squares support vector machine(LS-SVM).In order to predict SSC and pH of apple,gray level cooccurrence matrix(GLCM)was used to extract 8 texture features from the hyperspectral images of apple.Feature variables were extracted from the spectral data of pre-processed hyperspectral images,spectral and texture fusion data of hyperspectral images,and near-infrared spectral data by CARS,and predictive models were established by partial least squares regression(PLSR)and LS-SVM.The results showed that both NIR and hyperspectral imaging techniques could determine the shelf life of Golden Delicious apples.The optimal model was established by 1D+UVE+BP based on hyperspectral images,and the accuracy rate was 100％.The quantitative prediction models for SSC were established using a 1D+CARS+PLSR approach based on near-infrared spectroscopy,which demonstrated the most effective predictive performance.The correlation coefficient of the prediction set(Rp)and the root mean square error of prediction set(RMSEP)values were found to be 0.9323 and 0.4036,respectively.The SNV+CARS+LS-SVM model,utilizing near-infrared spectroscopy,demonstrated the most effective predictive performance,with Rp and RMSEP values of 0.8749 and 0.0417,respectively.The findings of this research offer valuable technical support and a foundational basis for the non-destructive testing of Golden Delicious apples.关键词
近红外光谱技术/高光谱成像系统/苹果/货架期/定性判别/定量预测Key words
near infrared spectroscopy/hyperspectral imaging system/apple/shelf life/qualitative discrimination/quantitative prediction分类
轻工纺织引用本文复制引用
赵昕,郑树亮,牛晓颖,曹建康,陈晗,赵志磊..基于高光谱成像技术和近红外光谱技术的金冠苹果货架期判别及其品质分析[J].食品工业科技,2025,46(11):302-312,11.基金项目
国家自然科学基金项目(32272371,32402230) （32272371,32402230）
2024年省级大学生创新训练计划项目(S202410075089). （S202410075089）
