摘要
A novel spectra-edge-matching approach is proposed for the quantitative analysis of the nonlinear fluorescence spectra of the air impurities excited by a femtosecond laser.The fluorescence spectra are first denoised and compressed,both by wavelet transform,and several peak groups are then picked from each spectrum according to a threshold of intensity and are used to extract the spectral features through principal component analysis.It is indicated that the first two principle components actually cover up to 98% of the total information and are sufficient for the final concentration analysis.The analysis reveals a monotone relationship between the spectra intensity and the concentration of the air impurities,suggesting that the femtosecond laser induced fluorescence spectroscopy along with the proposed spectra analysis method can become a powerful tool for monitoring environmental pollutants.
A novel spectra-edge-matching approach is proposed for the quantitative analysis of the nonlinear fluorescence spectra of the air impurities excited by a femtosecond laser.The fluorescence spectra are first denoised and compressed,both by wavelet transform,and several peak groups are then picked from each spectrum according to a threshold of intensity and are used to extract the spectral features through principal component analysis.It is indicated that the first two principle components actually cover up to 98% of the total information and are sufficient for the final concentration analysis.The analysis reveals a monotone relationship between the spectra intensity and the concentration of the air impurities,suggesting that the femtosecond laser induced fluorescence spectroscopy along with the proposed spectra analysis method can become a powerful tool for monitoring environmental pollutants.
基金
supported by the National Basic Research Program of China ("973" Program) (Grant No.2007CB310403)
the Key Program of the Natural Science Foundation of Tianjin (Grant No.07JCZDJC05500)
the National Natural Science Foundation of China (Grant Nos.60772105,10704043)
