Effect of wavelet transform techniques upon the estimation of sugar content in apple with near-infrared spectroscopy

Abstract

Wavelet transform (WT) has proven a powerful and efficient tool for dealing with chemical data due to its characteristic of dual localization and has been widely used in analytical chemistry. This paper aims at serving three purposes: First, it gives a review of the applications of the wavelet transform in infrared spectroscopy; Second, it gives a quick summary of aspects and properties of wavelets and wavelet transforms which are needed in order to understand how to (pre-) process data from spectrometry with wavelet methods; Third, it shows on a typical example (apple NIR spectra) how wavelet transforms can be used in order to extract quantitative information. The sugar content of intact apple was measured by NIRS and analyzed by wavelet transform, which is a new development in signal treatment method in recent years. The results show that the spectra treated with wavelet transform indicate more effectively the relationship with sugar content in intact apple. Compared with original spectra, wavelet transform of three-size has the most marked relation with sugar content. The predicting precision of five-element regression is the best and the scale 3 is the best for its 0.904 correlation efficient of determination and the 0.777 in standard error of prediction which is less than that of primitive spectra. Therefore, the conclusion of improved predicting precision for quantitative detection of sugar content in intact apple with wavelet transform can be drawn.