小波变换在岩屑样品红外漫反射光谱中的应用

测定了不同粒度烃源岩粉末的红外漫反射光谱。结果表明，样品原始光谱吸光度的测量值随着粒径的减小而减小；利用小波变换对原始光谱进行预处理，消除了由散射引起的背景和基线漂移后，发现光谱的吸光度测量值随着粒径的减小而增加。这说明粉末样品的散射对光谱线形的影响很大，因此在定量分析中应保持被测样品的粒度与建模时样品的粒度一致，以减小由散射差异所带来的误差。小波变换滤掉的光谱低频信息反映了样品的散射强弱，利用该低频信息在特征吸收峰2 820 cm-1附近建立预测样品粒径的模型，并与利用原始光谱信息建立的模型比较，前者相关性远高于后者，并且相关系数达到r=0.999 7，可用于预测和控制样品的粒度分布，从而提高红外光谱定量分析的精确度。同时文章还验证了散射系数S和吸收系数K都与粒径d成反比关系。

Application of Wavelet Transform to the Infrared Diffusion Reflectance Spectrum of Rocks

The infrared diffuse reflectance spectra of hydrocarbon source rocks with different particle sizes were measured. The result indicated that the absorbency of the raw spectrum decreased with the reduction of particle size, but the relationship turned to be reverse after we pretreated the original spectra by using wavelet transform to eliminate the background and calibrate the baseline drift, both of which were caused by scattering. The reversed relations showed that the spectral lines were influenced deeply by the scattering of the samples. So the particle size of the samples to be measured and the particle size of the model samples must be consistent to reduce the error. The low frequency part of the spectrum filtrated by wavelet transform corresponds to the scattering, and the authors used it to set up a model at the latent absorbance wave number (near 2 820 cm-1) to forecast the particle size. By comparing this model with the other model based on the original spectrum the authors found that the pertinence of the anterior model is higher than the latter one and the value reaches 0.999 7. So the authors can accurately forecast and control the distribution of the particle size by this model, which can be used to improve the accuracy in the quantitative analysis of the infrared reflectance spectrum. Also the study validated that both scattering and absorption coefficients are inversely correlated with the particle size.