Influence of Data preprocessing on near infrared (NIR) spectroscopy calibrations: A simulation study

Li Xing,Fengmin Zhao,Youfu Cao,Longjian Chen

2013-01-01

Abstract:Data preprocessing played an important role in near infrared (NIR) spectroscopy study. In this study, different preprocessing methods have been applied to two different datasets: NIR spectra of wheat and juice. In addition to the most common preprocessing methods such as derivatives, multiplicative scatter correction, Savitzky-Golay smoothing and standard normal variate, the direct orthogonal signal correction (DOSC) which is a kind of new preprocessing method was also tested. The NIR calibration models for two datasets were developed by using partial least squares method. The results showed that common preprocessing methods could not achieve consistent improvements for the two datasets. In contrast, DOSC method was found to be an excellent method for correcting NIR spectra. The use of DOSC method could not only consistently improve the performances of NIR calibration models for the two datasets, but also simplify NIR calibration models by concentrating the relevant information in fewer latent variables.