An Improved Background

In the traditional background correction algorithm based on the wavelet transform, approximation coefficients considered as frequency responses of background signal are usually set to zero. However, there are many meaningless negative values generated in the background-corrected spectrum because of the calibration errors of this algorithm. Intensities of some weak peaks even become negative and these peaks will disappear after the calibration of negative values. To solve these problems for the background correction of Raman spectrum, an improved intelligent algorithm which utilizes a suppression coefficient to modify approximation coefficients is proposed in this paper. A series of simulation analyses, as well as experimental investigations, are made to test the performance of this algorithm. It is proved that the use of the suppression coefficient could increase the background correction accuracy and decrease the number of meaningless negative values in the reconstructed spectra, which will prevent the disappearance of weak Raman peaks after the calibration of negative values and increase the sensitivity of Raman spectral analysis