基于主成分和多类判别分析的可见-红外光谱水蜜桃品种鉴别新方法

提出了一种用可见-近红外漫反射光谱技术快速鉴别水蜜桃品种的新方法.应用可见-近红外光谱仪测定三个品种水蜜桃的光谱曲线,再用主成分分析法对不同品种样本进行聚类分析,获取了水蜜桃可见-近红外光谱的特征信息,同时结合多类判别分析技术建立水蜜桃品种鉴别的模型.对经过预处理的光谱数据进行主成分分析,分析表明,以样本在第一主成分和第二主成分上的得分做出的二维散点图,对不同种类水蜜桃具有很好的聚类,能定性区分不同种类水蜜桃;经过主成分分析得到的前8个主成分的累积可信度已达94.38%,说明这8个变量能够代表绝大部分原始光谱的信息.从75个样本中随机抽取60个样本用于建立8个主成分变量的多类判别分析品种鉴别模型,余下的15个样本用于验证,准确率为100%.说明本文提出的方法具有明显的分类和鉴别作用.

NEW APPROACH OF DISCRIMINATION OF VARIETIES OF JUICY PEACH BY NEAR INFRARED SPECTRA BASED ON PCA AND MDA MODEL

A new method for discrimination of varieties of juicy peach by means of visible-near infrared spectroscopy(NIRS) was developed.First,the spectral curves of three varieties juicy peaches were measured by spectrometer;the pretreated spectra data of juicy peach were analyzed through principal component analysis(PCA).Then the diagnostic information from PCA was used as inputs of multiple discriminant analysis(MDA) for pattern recognition.The 2-dimontional plot was drawn with first and second principal components,which indicated that it was a good clustering analysis for classification varieties of juicy peach.The result of the analysis suggested that the reliabilities of first 8 principal components were more than 94.38%.60 samples from three varieties selected randomly.Then they were used to build discriminating model.15 unknown samples were validated by this model.The recognition rate is 100%.This model is reliable and practicable.So this study can offer a new approach to the fast discrimination of varieties of juicy peach.