基于小波分量奇异值分解的单通道盲分离算法.pdf

文档介绍：第 25 卷第 11 期电子测量与仪器学报 Vol. 25 No. 11
2011 年 11 月 JOURNAL OF ELECTRONIC MEASUREMENT AND INSTRUMENT · 991 ·

DOI: .
基于小波分量奇异值分解的单通道盲分离算法*
张纯1,2 杨俊安1,2 叶丰1
(1. 解放军电子工程学院, 合肥 230037; 2. 安徽省电子制约技术重点实验室, 合肥 230037)
摘要: 针对单通道盲分离领域先验信息不足的问题, 提出了一种基于小波变换、奇异值分解和独立分量分析的单通道盲
分离算法。首先对单路传感器接收的信号进行小波分解和重构, 得到一组代表原始信号特征的细节信号, 对其施加奇异值分解
并剔除小于门限值的奇异点, 以此消除干扰信号和噪声的影响。然后将经过处理的细节信号组作为独立分量分析算法的输入信
号, 通过独立分量分析算法来恢复原始信号。实验信号采取仿真信号和实际信号。实验结果表明, 该算法不需任何先验信息、
鲁棒性强、运行速度快、分离效果优异。
关键词: 单通道; 盲分离; 小波变换; 奇异值分解; 独立分量分析
中图分类号: 文献标识码: A 国家标准学科分类代码:
Single channel blind separation algorithm based on singular value
position of ponents
Zhang Chun1,2, Yang Jun’an1,2, Ye Feng1
(1. Electronic Engineering Institute of PLA, Hefei 230037, China; 2. Anhui Key Laboratory of Electronic Restriction,
Hefei 230037, China)
Abstract: Aiming at the problem of prior information deficiency in single channel blind separation field, a novel
method based on wavelet transformation, singular value position (SVD) and ponent analysis (ICA)
is proposed. Firstly the wavelet position and reconstruction is applied on the single channel mixed signal and a set
of detail signals which represents the feature of source signals are obtained. Then SVD is adopted on this set of signals and
values, which is less than threshold, is rejected. Therefore, the in