生物气溶胶红外光谱信号预处理算法研究

Research on Preprocessing Algorithm for Infrared Spectral Signals of Biological Aerosols

被动傅里叶变换红外(FTIR)遥感是一种具有应用潜力的生物气溶胶远程探测技术。红外遥感测量中目标光谱特征上往往存在噪声信号和基线漂移。而生物气溶胶的光谱特征相对较宽,传统的基线校正方法都不适用。由于生物气溶胶红外光谱和不同形式的基线漂移都是非高斯信号,把非高斯性作为独立性度量,基于独立成分分析(ICA)技术设计了生物气溶胶红外光谱信号的预处理算法。试验结果表明,该算法可以把未知干扰成分、基线漂移等作为独立分量分离出来,从而不影响进一步的定性、定量分析。

It has the application potentials of standoff detection of biological aerosol by passive infrared remote sensing based on Fourier-transform infrared （FTIR） spectroscopic technique. There are often noise signals and baseline drift on the target spectral signature in infrared remote sensing measurement. The spectrum of biological aerosol is relatively broad and the traditional methods of baseline correction are inapplicable. Due to infrared spectra of biological aerosol and different baseline drift are non-Gaussian signals, an algorithm for preprocessing infrared spectra of biological aerosol is devised based on independent component analysis （ICA）, where non-Gaussian is used as independent measure. The results of experiments show that this algorithm can separate unknown interference and baseline drift as independent component, and with no effects on the further qualitative and quantitative analysis.