基于主成分分析与独立成分分析的热释电红外信号特征提取技术

摘要

In this paper ,a new way has been presented to extract the character of pyroelectronic infrared sensor ( PIR ) signals from human and mice by principal component analysis ( PCA ) and independent component analysis (ICA).Firstly,the noses is removed from the original signals .Then the processed signals is translated from time domain to frequency domain by FFT .Next the signals are compressed by PCA in order to select the main useful information which remains the 99.95%of the original information . In the end , the principal typical components that is measured by kurtosis are picked up precisely from human and mice by the use of ICA .Experimental data show that the principal typical components can efficiently express the huge difference between the feature information about human and mice .In brief, this method provides an effective and feasible approach for obtaining the character of PIR signals from human and mice .%提出一种基于主成分分析（ PCA）和独立成分分析（ ICA）的人与老鼠热释电传感器红外信号特征提取的方法。首先对采集到传感器数据进行去噪预处理，并使用FFT变换到频域分析。然后用主成分分析法提取频谱数据主要信息，降低数据冗余量，同时保留了99．95％以上原始信息。最后使用独立成分分析法提取统计独立的独立成分，并用峰度系数来描述人和老鼠的独立特征分量信息。实验结果说明：提取的特征量都能充分描述人与老鼠的实际信息，且人与老鼠的特征差异足够明显，为提取人和老鼠热释电红外信号的特征提供一个有效可行的方法。