基于相关矩阵的频域盲分离排序方法

摘要

Ambiguities in permutation and amplitude are the main problems in frequency domain blind signal separation.A fast and efficient frequency domain method of solving the permutation ambiguity for blind source separation is investigated in order to solve the permutation ambiguity.A correlation matrix between two neighbor frequency bins is developed.It is transformed into a permutation matrix to deal with the permutation ambiguity.it’s simulation results illustrate that the new method improves average SNR and cuts down computation greatly.%排列模糊性和幅度模糊性一直是在频域上分离卷积混合信号所面临的主要问题,针对该问题,给出了一种基于相邻频点的幅度相关之和的、快速有效地解决频域分离算法中排列模糊性的方法;即通过定义相邻频点的相关矩阵并通过其置换形式来解决排列模糊问题,从而在频域上有效的分离源信号,仿真证明,该算法可以对卷积混合信号实行快速有效分离,分离效果理想,并大幅减少计算量。