The Handling Approach of Near-Infrared Spectroscopy for Apple Quality Prediction Based on Digital Signal Processing

Spectral preprocessing and spectral screening are important in spectroscopic analysis. This paper uses near-infrared diffuse reflectance technology to test Soluble Solid Content (SSC) of Red Fuji apple. In order to study the effect of spectral preprocessing and variable selection on the accuracy of SSC prediction model in this paper, we use the wavelet packet threshold de-noising, Savitzky-Golay smoothing and Multivariate Scatter Correction (MSC) to preprocess the spectrum, then evaluate the method of equal-interval decimation variable screening and a new wavelength screening method based on Wavelet Packet Analysis (WPA). The results show that the Partial Least Squares (PLS) model has better performance by adopting the spectra pretreated by the combination of wavelet packet threshold de-noising and MSC and filtered by WPA.