A Method of Distinguishing Tea varieties Based on Hyperspectral Imaging

Abstract

In order to realize the rapid and non-destructive identification of tea varieties, this paper based on hyperspectral imaging technology to find the optimal discrimination model of tea varieties. This article is mainly divided into three aspects: the discriminant model of tea varieties based on spectral characteristics, the discriminant model of tea varieties based on image features, and the discriminant model of tea varieties based on spectral-image fusion features. The experimental results show that Model 1 uses the full-spectrum feature combined with support vector machine (SVM) model, which can distinguish the accuracy of different tea varieties up to 100%. Model 2 is based on the GLCM texture feature based on the characteristic gray image combined with the SVM model, and the discrimination accuracy of different tea varieties reaches 100%. Model 3 discusses the impact of different preprocessing methods on the accuracy of classification under the fusion of two information features, determines Minmax as the best preprocessing method, and obtains 100% classification accuracy in the test set.