The method research on nondestructive measuring of vitamin C content in orange

In order to explore a method for nondestructive measuring of vitamin C content in orange, the effects of 11 pretreatment methods on the prediction performance by the model based on PLS have been compared in this paper. It is shown that, among them, the wavelet transform can give the best denoising effect. In addition, a comparison of the denoising effects produced by Daubechies3 wavelet transform at different decomposing leves was also made. The results show that the prediction performance by PLS modeling varies with the different wavelet decomposing levels and level 4 gives the best result, with the correlation coefficient R between the real and predicted values reaching 0.9574 and RMSECV being only 3.9mg/100g. It is also found that the optimal spectral wave band best reflecting the Vc content of orange is between 7501.7 cm-1 and 5449.8 cm-1, and the best principal component for PLS models is 8. In conclusion, the modeling based on PLS with near-infrared spectroscopy followed by wavelet denoising is an effective method to nondestructively measure vitamin C content of orange.