基于复数小波的X射线荧光光谱本底扣除法

A background removing approach based on complex wavelet transform for X-ray fluorescence spectrometry

在X射线荧光光谱(XRF)分析中,本底能否有效扣除对分析结果的精确性有很大影响。针对实数小波变换系数在奇点附近有正负振荡及缺乏平移不变性等问题,提出了一种使用近似解析的复数小波扣除本底的新方法。该复数小波变换通过两个实数小波来实现,所以方法兼有实数小波局部时频分析和多分辨率等优点。采用复数小波分解光谱后,利用低频逼近重构信号得到本底。为了检验其有效性,对一个模拟的光谱和实验测得的光谱进行了本底扣除,并通过分析比较了方法对特征峰净峰面积和峰位置的影响,实验结果显示该复数小波得到的本底比一般实数小波的本底更为精确,证明了方法可以有效扣除本底并能保持峰面积等有效信号不变。

In X-ray fluorescence spectrometry(XRF)analysis,whether the background could be removed validly has considerable effect on the analysis accuracy.Because of the oscillation of real wavelet coefficients around the singularities and lack of shift-invariance,a novel background removing method based on approximately analytic complex wavelet transform is presented.The complex wavelet is implemented through two real wavelets,so it has real wavelet's main properties that are time-frequency localization and multi-resolution analysis.The spectrum is decomposed with the complex wavelet,and then the background is obtained through the low frequency approximation.In order to test its effectiveness,the algorithm is used to extract the backgrounds of one simulated spectrum and one experimental spectrum.The effect to the desired information,such as net peak areas and peak positions,is also considered.Experiments results show the backgrounds obtained through the proposed complex wavelet transform are more accurate than common real wavelet transform.A conclusion could be drawn that the proposed algorithm could remove the background accurately while keeping the desired information intact.