基于近红外技术和偏最小二乘判别法对无花果成熟度的快速判别

Identification of fig maturity based on near-infrared spectroscopy and partial least square-discriminant analysis

以无花果为试验对象,对其进行近红外光谱采集,并对其糖度、单果重、纵径、横径、硬度5个指标进行K-均值聚类;根据光谱数据、主成分分析确定最优聚类效果的成分和各类别的指标分布构建偏最小二乘判别分析(PLS-DA)模型进行聚类判别,以实现对果实成熟度(幼果期、成长期、成熟期)分类的准确、快速、无损伤鉴别。结果表明,3种成熟阶段的无花果样品的糖度、单果重和硬度均具有显著性差异,成熟果和成长果与幼果的纵径和横径间具有显著性差异。根据PLS-DA判别模型累计训练集的分类正确率为99.59%,测试集的分类正确率为99.15%。说明主成分分析与光谱数据所建立的PLS-DA模型性能较好,对无花果成熟度的快速鉴别是有效且可行的。

In order to distinguish figs of different maturity accurately and quickly without damage,the samples of figs were collected by near-infrared spectroscopy,and K-means clustering was carried out for five indexes of fig sugar degree,single fruit weight,vertical diameter,horizontal diameter and hardness.According to the spectral data and the score chart of principal component analysis,the components of the optimal clustering effect and the index distribution of each category were determined.Based on the PLS-DA model,the cluster discrimination model was constructed to achieve the purpose of fruit maturity classification.Through the analysis of the differences of the above five indexes,significant differences in soluble solids,single fruit weight and hardness of the three kinds of fig samples at the maturity stage were found,and significant differences between the vertical and horizontal diameters of mature and growing fruits and young fruits were also detected.According to PLS-DA discriminant model,the classification accuracy of training set was99.59%,and that of test set was 99.15%.The results showed that the PLS-DA model based on principal component analysis and spectral data had good performance,and it could be used to identify the maturity of figs quickly.