基于快速独立分量分析的多维混合信号盲分离  

Blind separation of multidimensional mixed signals based on fast independent component analysis

作　　者：吴正茂[1] 孙云莲[1] 

机构地区：[1]武汉大学电气工程学院

出　　处：《武汉大学学报（工学版）》2005年第4期38-41,共4页Engineering Journal of Wuhan University

摘　　要：介绍了一种用于盲信号分离的快速独立分量分析方法(FastICA).该方法以信息论中的负熵作为估计输出分量之间统计独立的目标函数,通过优化该目标函数,得到快速独立分量分析的迭代算法.由于该迭代算法不需计算输出分量的高阶统计量,收敛速度快.通过线性混合时间信号分离和图像信号分离的仿真实验表明Fas-tICA算法可以快速有效地分离这些信号.A new fast algorithm of independent component analysis（FastlCA） applied to blind signal separation is introduced. In the method, negentropy of information theory is used as the objective function which estimates statistical independence between output components . A fast iterative algorithm is derived by optimizing the objective function. The method does not need to calculate the higher order statistics of output components and converges fast. The simulation experiment on the separation of linear mixed time signals and image signals are done by FastICA algorithm. The efficiency of the FastICA algorithm is verified by the simulation experiment results.

关 键 词：盲信号分离 独立分量分析 FASTICA算法 

分 类 号：TN911.17[电子电信—通信与信息系统]