Calibration Transfer of Near-Infrared Spectroscopic Model for Soluble Solid Content Predication of Apples by the Combined Use of Direct Standardization and Piecewise Direct Standardization

To achieve the transfer of a near-infrared (NIR) calibration model for determining the soluble solid content (SSC) in apples, this study employed the spectral data and SSC values of 89 apple samples measured with two portable NIR spectrometers. After outlier removal and preprocessing, the calibration set from the master instrument was used to establish the NIR model by partial least-squares regression (PLSR). Subsequently, the transfer of the NIR model between the master and slave instruments was accomplished by the combined use of direct standardization (DS) and piecewise direct standardization (PDS). The results indicated that compared with DS and PDS, the combined algorithm not only significantly improved the prediction performance of the model for the slave instrument, but also mitigated the artifacts caused by PDS. After calibration transfer, the ratio of standard deviation of the validation set to standard error of prediction (RPD) increased from 1.585 3 to 3.264 5, while the root mean square error of prediction (RMSEP) decreased from 1.093 2 to 0.530 9. Therefore, the proposed DS-PDS algorithm leverages the advantages of DS and PDS, allowing successful transfer of the calibration model between the two portable spectrometers.