PCA和SPA的近红外光谱识别白菜种子品种研究

发布时间：2018-03-28 16:10

本文选题：近红外光谱　切入点：主成分分析　出处：《光谱学与光谱分析》2016年11期

【摘要】：为了实现对不同品种白菜种子的快速无损鉴别,应用近红外光谱技术获取白菜种子的光谱反射率,首先采用变量标准化校正和多元散射校正对原始光谱进行预处理;其次,采用主成分分析法(PCA)对光谱数据进行聚类分析,从定性分析的角度得到三种不同白菜种子的特征差异,并采用连续投影算法(SPA)选取特征波长;最后,分别基于全波段光谱、PCA分析得到的前3个主成分变量以及SPA算法选取的特征波长,建立了最小二乘支持向量机(LS-SVM)和偏最小二乘判别(PLS-DA)模型进行白菜种子不同品种的鉴别。从主成分PC1、PC2得分图中可以看出,主成分1和2对不同种类白菜种子具有很好的聚类作用。基于特征波长建立的PLS-DA和LS-SVM模型的判别结果优于基于主成分变量建立的模型,其中基于特征波长建立的LS-SVM模型识别效果最优,建模集和预测集的品种识别率均达到100%。结果表明,通过SPA算法选取的6个特征波长变量能够很好的反映光谱信息,提出的SPA算法结合LS-SVM预测模型能获得满意的分类结果,为白菜种子品种的识别提供了一种新方法。
[Abstract]:In order to obtain the spectral reflectance of different varieties of Chinese cabbage seeds by near infrared spectroscopy, the original spectrum was pretreated by variable standardization correction and multivariate scattering correction. The spectral data were clustered by principal component analysis (PCA), and the characteristic differences of three different kinds of Chinese cabbage seeds were obtained from qualitative analysis, and the characteristic wavelengths were selected by continuous projection algorithm (spa). Based on the first three principal component variables and the characteristic wavelength selected by SPA algorithm, the first three principal component variables are obtained from full-band spectral analysis. The least squares support vector machine (LS-SVM) and partial least squares discriminant (PLS-DAA) models were established to identify different varieties of Chinese cabbage seeds. Principal component 1 and 2 have a good clustering effect on different species of Chinese cabbage seeds. The discriminant results of PLS-DA and LS-SVM models based on characteristic wavelength are better than those based on principal component variables. Among them, the LS-SVM model based on characteristic wavelength is the best, and the recognition rate of the model set and prediction set is 100. The result shows that the six characteristic wavelength variables selected by SPA algorithm can reflect the spectral information very well. The proposed SPA algorithm combined with the LS-SVM prediction model can obtain satisfactory classification results and provide a new method for the identification of Chinese cabbage seed varieties.
【作者单位】：华东交通大学;
【基金】：国家自然科学基金项目(61565005)资助
【分类号】：S634;O657.33

【相似文献】