不同产地稻米的二维相关近红外光谱鉴别

Identification of Rice from Different Origins by Two-dimensional Correlation Near Infrared Spectroscopy

为了实现针对不同产地稻米样品的无损判别,提出一种基于同步二维相关近红外光谱的欧氏距离判别方法。在分别得到5种稻米样品的标准二维相关谱后,通过计算测试集样品与各个标准相关谱的欧氏距离,根据最小距离原则进行归类判别,经实际测试正确率达到95%。通过定义区分度系数还可以更精确判断对各类别样品的区分效果。为了比较二维相关方法与常用建模方法的实际效果,在同样的数据基础上进行偏最小二乘判别分析和支持向量机分类的尝试,其最高正确率分别为75%和70%,验证了二维相关方法的有效性。本研究为无损鉴别稻米样品提供了一种新的可能方法,并可为在线检测提供参考。

In order to realize the nondestructive discrimination of rice samples from different origins,a discrimination method based on synchronous two-dimensional correlation near infrared spectroscopy and Euclidean distance was proposed.After obtaining the standard two-dimensional correlation spectra of five types of rice samples respectively,the Euclidean distance between the test set samples and each standard correlation spectrum is calculated,and the discrimination is based on the minimum distance.The accuracy of this method can reach 95%.By defining the discrimination coefficient,the discrimination effect of various types of samples can also be judged more accurately.In order to compare the practical effects of two-dimensional correlation method and common modeling methods,partial least squares discriminant analysis and support vector machine classification are tried on the basis of the same data.The highest accuracy rates are 75%and 70%respectively,which verifies the effectiveness of two-dimensional correlation method.This study provides a possible method for nondestructive identification of rice samples,and can provide a reference for on-line detection.