摘要
该文针对源信号时域和频域不充分稀疏的情况,提出了欠定盲源分离中估计混合矩阵的一种新方法。该方法对等间隔分段的观测信号应用独立分量分析(ICA)的盲分离算法获得多个子混合矩阵,然后对其分选剔除了不属于原混合矩阵的元素,最后利用C均值聚类的学习算法获得对混合矩阵的精确估计,解决了源信号在时域和频域不充分稀疏的情况下准确估计混合矩阵的问题。在估计出混合矩阵的基础上,利用基于稀疏分解的统计量算法分离出源信号。由仿真结果,以及与传统的K均值聚类,时域检索平均算法对比的实验结果说明了该文算法的有效性和鲁棒性。
A method of the mixing matrix estimation in the underdetermined source separation is proposed in which the sources are not sparse enough to estimate the mixing matrix. Getting many sub matrixes through applying Independent component analysis(ICA) for observation signals and removing the elements do not belong to the mixing matrix,the mixing matrix is estimated precisely with C-means clustering agglomeration. Then,the source signals can be recovered with the statistically sparse decomposition principle. The experiment shows that the method have better accuracy and validity than K-means and searching-and-averaging method in the time domain in estimating the mixing matrix.
出处
《电子与信息学报》
EI
CSCD
北大核心
2010年第4期919-924,共6页
Journal of Electronics & Information Technology
关键词
信号处理
欠定盲源分离
独立分量分析
C均值聚类
稀疏分解的统计量
Signal processing
Underdetermined blind signal source separation
Independent Component Analysis(ICA)
C-means clustering
Statistically sparse decomposition principle