A novel method in spectroscopy signal processing

Zhang Lu,Zhao Hong,Wang Xiaopin

DOI: https://doi.org/10.1117/12.754171

2007-01-01

Abstract:Many methods for spectroscopy signal analysis have been employed to extract useful and correct signals from measured data, such as Fourier analysis, wavelet analysis and so on. However, Fourier analysis is only suitable for stationary signal processing. Although wavelet transform is capable of analyzing non-stationary signals, many deficiencies have been reported in the use of wavelet transform. Non-adaptive nature is one of its disadvantages. Once the basic wavelet is selected, one will have to use it to analyze all the data. And the number of levers that the signal will be decomposed into must be decided by user and the frequency bands of all lever signals are fixed. Due to the deficiencies of wavelet transform, Huang et al. proposed a new type of signal processing method called empirical mode decomposition (EMD), with which any complicated data set can be decomposed into a finite and often small number of Intrinsic Mode Functions (IMFs). This method is suitable for non-linear and non-stationary data processing and has advantages that wavelet analysis has. In this paper EMD method is used in spectroscopy signal processing. And it is just in the infant period for spectroscopy signal processing. In order to verify the rationality and feasibility of EMD, polystyrene thin film is chosen as the testing sample. Its reflection interference spectroscopy is collected by the spectrometer made by Ocean Optics Company. And then the EMD approach for processing the reflection interference spectrum of polystyrene film is discussed. The sifting process in EMD can be stopped by predetermined criteria. And the number of IMF is self-adaptive. For choosing suitable number of IMF, the variance standard has been selected in this paper, with which the quality of the processing results can be optimized. Then the thickness of the polystyrene film can be calculated from the extracted spectrum by EMD. Comparing the thickness with the calibrated value, the error is below 1%, which proved that the proposed method is efficient and accurate in spectroscopy signal analysis.