Fast quantification of matcha adulterants with laser-induced breakdown spectroscopy spectrum and image.

Matcha adulteration is a critical issue in food production, which may reduce the effective components in matcha and have a detrimental effect on human health. Hence, laser-induced breakdown spectroscopy (LIBS) combined with chemometric methods was used to fast quantify the adulterant content. Two types of matcha adulteration, including mixing two grades of matcha (Fuding and Jiukeng) with barley seedling powder, were quantified with fusion of LIBS spectrum and ablation image. In addition, important features of spectrum and ablation image were extracted by extreme gradient boosting (XGBoost) and gray-level co-occurrence matrix (GLCM), respectively. Support vector machine regression models were established based on full spectrum, spectral features, and feature fusion. The models based on fusion of spectral and textural features achieved the best performance with R of 0.9046, RMSE of 0.12 for Fuding, and R of 0.9200, RMSE of 0.11 for Jiukeng. This study provides a fast approach for quantifying matcha adulteration.