利用近红外光谱技术识别不同类别的茶叶

以龙井、碧螺春、祁红和铁观音4种中国名茶为对象,研究了采用近红外光谱结合K最近邻法(KNN)模式识别方法对茶叶进行识别与分类的可行性.选取6500～5500 cm-1(1538～1818 nm)波数范围内的光谱,通过标准正态变量变换(SNV)预处理后,利用KNN的模式识别方法建立识别模型.结果表明,4主成分因子建立的KNN判别模型最佳,模型对训练集与预测集中样本的识别率都达到100%.该结论为快速准确识别茶叶提供了一种新思路.

Application of Near Infrared Reflectance Spectroscopy in Identification of Different Classified Tea

With 4 kinds of famous tea of Longjing,Biluochun,Qi and Tieguanyin as test materials,the feasibility of identification and classify of tea was studied by using near infrared reflectance spectroscopy coupled with pattern recognition based on KNN.It was found that in the spectra range between 6 500 cm-1 and 5 300 cm-1,the models of tea variety identification were built separately with KNN pattern recognition method after the pretreatment by standard normal variate transformation(SNV).Results showed that the KNN identification models were optimal when 4 principal components factors were used in building models,with the identification being all 100 % in samples of training set and forecast set.This conclusion offered a new idea for the quick and precise identification of tea.