摘要
简要阐述了独立成分分析(independent component analysis,ICA)的基本模型及其假设、含混性、非高斯性度量和通用求解过程,介绍了一种基于峰度的快速ICA算法。提出了基于基本ICA模型的从被动遥感红外光谱中分离出弱目标信号的信号检测方法。实验结果表明:基于ICA的信号提取方法可不依赖于预先采集的"干净"背景光谱,并且与差谱法的结果进行了对比。
The standard model of independent component analysis （ICA） and its assumptions, ambiguities,nongaussianity measures and general solution were introduced. A kind of fast ICA algorithm based on kurtosis was discussed. Then, A weak signal feature extraction algorithm for passive infrared spectra based on standard model of ICA was proposed. Comparison was made with difference spectra. Experimental results show that the proposed algorithm can effectively detect the weak signal form passive infrared spectra,and doesn＇t depend on background spectra.
出处
《激光与红外》
CAS
CSCD
北大核心
2008年第3期289-291,299,共4页
Laser & Infrared
关键词
独立成分分析
被动红外光谱
信号
检测
independent component analysis
passive infrared spectrum
signal
detection
