基于透射光谱的沙田柚粒化分级及综合品质检测

Comprehensive Quality Detection of Shatian Pomelo in Different Granulation Levels Based on Transmission Spectroscopy

汁胞粒化是柚果内出现汁液囊变硬、干燥等特征的生理性病害。沙田柚皮厚果大,在采后贮藏期发生汁胞粒化,对内部品质及口感影响较大。文章采用一种可见-近红外透射光谱技术,用于沙田柚粒化分级和综合内部品质检测。沙田柚挂果成熟后,采摘600个样本用于实验,分别采集400~1100 nm与900~1700 nm波段光谱,经光谱预处理和特征提取,结合化学计量学和“深度学习”模型,探究柚果按粒化程度分为5级后,其贮藏期内部品质变化规律。根据理化指标构建沙田柚综合品质指标,并建立支持向量机回归模型,其预测集决定系数和均方根误差分别为0.9481和4.8343;与此同时,建立SCARS-SVM-DA和SCARS-LDA模型,采用混淆矩阵学习和模型评价体系用于汁胞粒化等级分类和预测。结果表明,该方法在汁胞粒化等级分类和综合品质指标检测中优势明显,以期为厚皮水果内部品质快速检测及病害监测提供参考和理论依据。

Granulation is a physiological disorder,wherein juice sacs become hard and dry,which damages the internal quality of pomelo.The granulation of Shatian pomelo with thick skin and large fruit shape occurred in the storage period after harvest,which had a profound influence on its internal quality and taste.In this study,a rapid and non-destructive detection method based on visible and near-infrared transmittance spectroscopy was used to detect granulation levels and comprehensive quality index in pomelo.600 samples were harvested and used in the experiment.The original spectra data of samples was recorded in the range of 400-1100 nm and 900-1700 nm.It was preprocessed and feature extraction,combined with the chemometrics method and deep learning models,to explore the internal quality change rules of granulated samples classified to the five granulation levels.According to the physicochemical indexes,the comprehensive quality index of Shatian pomelo was constructed,and the support vector machine regression model was established.The model prediction set coefficient of determination and root the means square error is 0.9481 and 4.8343,respectively.Meanwhile,the SCARS-SVM-DA and SCARS-LDA models were established,and the confusion matrix learning and model evaluation system were used for the classification and prediction of granulation levels.The results show that this method has obvious advantages in the classification of pomelo in different granulation levels and the detection of comprehensive quality indicators.So it can also be utilized to research the internal quality and disease of thick-skinned fruit in rapid nondestructive testing.