Nondestructive Examination of VC Content of Intact Shatangju with Near Infrared (NIR) Spectroscopy Based on Wavelet De-noising

摘要

The objective of this research was to determine how NIR measurements of VC content in Shatangju (Citrus reticulate Blanco) was affected by wavelet de-noising (WD) method. Firstly, outlier samples (NO.3, 17, 76) were removed based on sample residuals. Then the spectra of 85 samples within 500-2500nm were decomposed in level 2,3 and 4 using the orthogonal wavelet functions DB n(n=2,3,4,5,6,7,8). Lastly the precision and stability of PLS model with different pretreatments were compared. WD was examined to be the optimal spectrum preprocessing method. The PLS model with DB2 wavelet de-noise (in decomposition level 4) produced best result (RMSEC=6.006, RMSECV=8.080, r_c =0.896 and r_v=0.808). The results showed that the NIR model treated by WD is feasible to detect VC content of Shatangju rapidly and nondestructively.