摘要: |
为了有效估计混合
矩阵并恢复出源信号,考虑到现实中的很多信号都是带限信号,提出了采用互补滤波器
组进行频带分解的欠定盲源分离方法。该方法将接收的混合信号经互补滤波器组分离到不同
的子频带,然后在每一个子频带分别估计混合矩阵进行常规的盲分离,利用聚类分析方法估
计总的混合矩阵,最后把相关的分离子频带信号进行叠加以恢复出源信号。即使源信号数目
远大于观测信号数,但只要保证在每一个子频带的信号数不大于观测信号数,该方法也可
以有效地分离出源信号。最后通过仿真验证了该方法的有效性及良好的分离性能。 |
关键词: 欠定盲源分离 互补滤波器 频带分解 聚类分析 |
DOI: |
|
基金项目:部级预研基金项目资助(9140A220309090C0201) |
|
Frequency Decomposition Method for Under determined Blind Source Separation by Complementary Filters |
FENG Tao,ZHU Li-dong |
(National Key Laboratory of Science and Technology on Communications, University
of Electronic Science and Technology of China,Chengdu 611731,China) |
Abstract: |
In view of the fact that many signals are bandlimited signals, a
new underdetermined blind source separation(BSS) method is proposed using fre
quency decomposition metho
d by complementary filters to estimate mixing matrix and recover source signals
effectively. The sensing signals are divided into different subbands, and each
subband uses common BSS algorithm.The source signals are constructed by co
m
bing the related signals together. Even if the number of source signals is much
larger than that of sensing signals, this method can also recover the sou
rce signals when the number of source signals is not more than that of sensing
signals in each subband. Computer simulation confirms the validity and excelle
nt separation performance of
this method. |
Key words: blind source separation complementary filters frequency decomposition cluster a
nalysis |