摘要
给出了欠定情况下的盲分离模型,从时域和频域对信号的稀疏特性做了简要的陈述。基于信号稀疏特性的欠定情况下盲分离一般采用两步法,对其核心步骤混叠分离矩阵A的估计中的聚类方法做了总结、归纳。对比了几种主要聚类方法的优缺点,并对今后欠定情况下混叠矩阵A的研究方向做了进一步展望。
The model of underdetermined blind signal separation（BSS） is put forward,and the sparsity of signals on time domain and frequency domain are stated briefly.Underdetermined BSS based on sparsity generally takes two-step approach,for the core of two-step,the estimation of mixing matrix,several clustering algorithms are summed up in this paper.The merits and drawbacks are compared among these algorithms,for the further research on mixing matrix in the future.
出处
《机械工程与自动化》
2010年第6期56-58,共3页
Mechanical Engineering & Automation
关键词
欠定盲分离
稀疏特性
聚类算法
underdetermined BSS
sparsity
clustering algorithms
