Abstract
A standardization algorithm, called WPTS, which is based on a wavelet packet transform (WPT) and entropy criteria, is proposed to transfer spectra between two near-infrared spectrometers. First, the spectra of the standardization samples measured on both spectrometers are collected and the difference mean spectrum is calculated; then, a wavelet packet transform is performed and a best WPT-tree is generated, of which a node is selected to establish a transfer matrix. Once the transfer matrix has been built, spectra measured on one spectrometer can be successfully transferred to another spectrometer as if they have been measured directly on the latter. By comparison, the proposed WPTS can reach a transfer performance comparable to WTS1, which is the best existing method, and better than WTS2 and classical PDS by a simpler technique of variable selection and lower computational cost of calibration transfer.
