基于高光谱成像技术的金银花与山银花快速鉴别

利用高光谱成像技术,研究一种快速、准确、无损的鉴别金银花与山银花的方法。通过对比3种预处理方法对偏最小二乘算法(Partial Least Squares,PLS)建模效果的影响,得到SNV为建模最优预处理方法。使用回归系数法(Regression Coefficient,RC)和连续投影算法(Successive Projection Algorithm,SPA)选择经预处理后光谱的特征波长,并分别建立极限学习机(Extreme learning machine,ELM)和最小二乘支持向量机(Last Squares Support Vector Machine,LSSVM)的判别分析模型。结果表明,光谱经SNV预处理后,应用SPA提取特征波长并建立LS-SVM判别分析模型为金银花和山银花最优判别模型,其建模集与预测集识别率均达到了100.00%。因此,利用高光谱成像技术能够无损、有效地鉴别金银花与山银花,并且在全光谱和特征波长下均能实现金银花与山银花的快速判别分析。

Rapid identification of Lonicerae Japonicae Flos and Lonicerae Flos based on hyperspectral imaging

In order to identify Lonicerae Japonicae Flos and Lonicerae Flos rapidly and precisely, a hyperspectral imaging technology combined with chemometric methods was applied to develop the nondestructive identification models for Lonicerae Japonicae Flos and Lonicerae Flos. Firstly, the original spectral data were analyzed by three pretreatment methods including Savitzky-Golay (SG) convolution smoothing, Multiple Scatter Correct (MSC) and Standard Normal Variate Transformation (SNV). A comparison was made among SG, MSC and SNV based on Partial Least Squares (PLS), of which the best pretreatment method was SNV. The Regression Coefficient (RC) and Successive Projection Algorithm (SPA) were used to extract the characteristic wavelengths after SNV pretreatment. Extreme learning machine (ELM) and Last Squares Support Vector Machine (LS-SVM) were applied to build the classification models based on characteristic wavelengths. This results revealed that the LS-SVM model based on SPA performed the optimal classification, with the accuracy of all 100% for modeling set and prediction set. Therefore, hyperspectral imaging technology can be used to identify Lonicerae Japonicae Flos and Lonicerae Flos effectively and non-destructively based on full wavelengths and characteristic wavelengths.