一种欠定盲源分离中混合矩阵的估计方法

A new method of mixing matrix estimation in underdetermined blind source separation

摘要:

欠定盲源分离问题中基于源信号稀疏性的两阶段法中，混合矩阵估计的准确与否，直接影响源信号的恢复效果。文中提出了一种在稀疏域估计混合矩阵的新方法。该方法通过搜索稀疏域中同一直线附近的点，利用这些点重构出混合矩阵，避免了远离直线周边的点对估计混合矩阵的干扰，从而大大降低了计算量。仿真表明该算法性能良好。

Abstract:

This paper focuses on the mixing matrix estimation stage，which is the first stage of the two-stage approach based on the sparse sources in underdetermined blind source separation. In the two-stage approach，the accuracy of estimating the mixing matrix will affect the recovered results to a great degree. A new method is put forward to estimate the mixing matrix in sparse domain. This method can recover the mixing matrix in sparse domain by searching the points next to the same line, thus can avoid the interference of the points far from the same line, and reduce the amount of computation greatly. Simulation results show that the method has a good performance.