Wavelet transform used to preprocessing of Apple near infrared spectra

摘要

Due to original near infrared (NIR) spectra data containing a lot of noise and large amount of data, so it is very necessary to do preprocessing for spectral data before spectral data analysis. Near infrared spectrum data preprocessing mainly includes two tasks: one is the noise reduction to improve the robustness of the model and the accuracy of the prediction results; second, data compression, in order to facilitate the storage of data, and improve the modeling speed. Traditional NIR spectral data preprocessing methods have limitations, and are difficult to get a satisfactory effect in both aspects. This paper will use wavelet analysis on preprocessing of Apple NIR spectral data, select peak signal-to-noise ratio (PSNR) and normalized correlation coefficient (NC) as evaluation indexes, and compared with the commonly-used Savitzky-Golay smoothing filter and multiple scattering correction, wavelet method can not only effectively realize the data compression, and also superior to the other two preprocessing algorithm in terms of noise reduction and spectral details maintain.