Non-destructive determination of volatile oil and moisture content and discrimination of geographical origins of Zanthoxylum bungeanum Maxim. by hyperspectral imaging

Abstract

Zanthoxylum bungeanum Maxim. is now known as a condiment type providing both medical applications and food. The volatile oil content and moisture content are two main quality indicators of Zanthoxylum, and the former is considered to be influenced by the growth region, climate and other environmental factors. Here, the feasibility of hyperspectral imaging in the range of 380–1040 nm to determine the volatile oil and moisture content of Z. bungeanum thus evaluating quality and distinguishing geographical origin were investigated. Regression statistical methods of partial least square, support vector regression (SVR) and extreme learning machine and wavelengths selection algorithms of competitive adaptive reweighted sampling (CARS) and variable combination population analysis were applied. The CARS-SVR models proved to be satisfying for predicting the volatile oil and moisture content. The correlation coefficient of calibration and residual predictive deviation values were 0.9059, 2.87 for volatile oil content and 0.8304, 1.94 for moisture content, respectively. Furthermore, SVM was used to classify the geographical origin of Z. bungeanum samples with an accuracy of 97.29% in the prediction set. The results suggested that HSI exhibits suitable characteristics for determining the volatile oil and moisture content in Z. bungeanum.