Apple near Infrared Spectral Preprocessed Based on Wavelet Analysis

The original near infrared spectral (NIR) inevitably includes noise signals and usually has a large amount of data. So de-noising and compression are the main tasks of spectral data pretreatment, in order to improve the accuracy of model prediction and modeling speed. Traditional spectral pretreatment methods have their own limitations in these two aspects. In this paper, the wavelet analysis method was applied to preprocess apple NIR data. Compared with Savitzky-Golay smoothing method and multiple scattering correction , the wavelet method has its own superiority in noise removal and data compression.