基于近红外光谱与矿物元素指纹信息融合对绿豆产地判别模型比较分析

Comparative Analysis of Mung Bean Discrimination Model Based on Near Infrared Spectrum and Fingerprint Information

为提高模型对绿豆产地的鉴别率,通过近红外光谱指纹信息和矿物元素指纹信息融合技术分析来自黑龙江省杜尔伯特蒙古自治县、吉林省白城市、黑龙江省泰来县、山东省泗水县绿豆样品中近红外光谱吸收强度和矿物元素含量,结合偏最小二乘判别分析(PLS-DA)法建立融合这两种指纹信息的鉴别方法。结果表明,信息融合模型的绿豆产地鉴别准确率为100%,与单一光谱指纹信息模型(90.0%)和矿物元素指纹信息模型(96.7%)相比,分别提高了10.0%和3.3%。因此,采用PLS-DA法信息融合模型对绿豆产地进行鉴别是可行的,近红外光谱指纹信息和矿物元素指纹信息融合技术可以提高绿豆产地的鉴别效果。

In order to improve the identification rate of mung bean origin in the model, in the present study, the near-infrared spectral absorption intensity and mineral element content of mung bean samples from Dulbert Mongolian Autonomous County of Heilongjiang Province, Baicheng City of Jilin Province, Tailai County of Heilongjiang Province, and Surabaya County of Shandong Province were analyzed by near-infrared spectral fingerprint information fusion technology, and the identification method of fusion of the two fingerprint information was established by combining the partial least squares discriminant analysis(PLS-DA) method. Theresults showed that the accuracy of mung bean origin identification in the information fusion model was 100%, which was 10.0% and 3.3% higher than that of the single spectral fingerprint information model(90.0%) and the mineral element fingerprint information model(96.7%), respectively. Hence, it is workable to identify mung bean producing areas by PLS-DA information fusion model. Moreover, near infrared spectrum fingerprint information and mineral element fingerprint information fusion technology can improve the identification effect of mung bean origin.