摘要：AbstractThe contents of neutral detergent fiber (NDF) and acid detergent fiber (ADF) in the tea stem are directly related to the mechanical properties of the stem, and consequently affects shearing force and shearing quality. Hyperspectral imaging of NDF and ADF might reflect their contents as they are organic matters with function groups. Therefore, shearing force of tea stems was determined by predicting NDF and ADF with hyperspectral imaging technique. The different planted time and 80 stem samples were collected from different tea plant with hyperspectral imaging technology, and partial least square regression (PLSR) were used to establish the models in this paper. The results showed that the calibration models with the spectral data preprocessing of the Savitzky-Golay smoothing and multiplicative scattering correction were the best for NDF and ADF. Determination coefficient of calibration and prediction models for NDF and ADF was higher than 0.8859, and ranged from 0.8859 to 0.9168 for cross and external validation. The root mean square errors of calibration (RMSEC) and root mean square errors of prediction (RMSEP) for NDF and ADF ranged from 2.676% to 4.161%. The relational model between shearing force and ADF was set up through stepwise regression, and the determination coefficient was 0.8110. The fitting coefficient between shearing force and reflectance was 0.9724. The study provides with a rapid method of determining shearing force of tea stem with hyperspectral imaging technique.
