基于源信号数目估计的欠定盲分离

Underdetermlned Blind Separation Based on Source Signals' Number Estimation

该文利用欠定盲分离下稀疏源信号的特点,估计源信号的数目且恢复源信号。通常在用两步法来解决欠定盲分离时,首先利用K-均值算法对观测信号聚类估计出混叠矩阵,最后利用最短路径法来恢复源信号,但是在以往的算法中,第1步估计混叠矩阵时,通常假设源信号数目是已知的,从而进行K.均值聚类,而事实上源信号数目根本无法知道,因此对源信号数目的估计对两步法有很重要的影响。因此本文提出了一种新的两步法算法,其中第1步利用稀疏源信号反映在观测信号中的特征来准确地估计出稀疏源信号的数目,且能得到混叠矩阵,从而恢复源信号。最后的仿真结果,以及与通常的K-均值聚类算法对比的仿真结果说明了此算法的可行性和优异的性能。

This paper gives a new method to estimate the number of source signals and recover them by the characteristics of sparse source signals in underdetermined blind separation. It is well known that source signals can be recovered through the two-step algorithms generally. The first step is to estimate the mixture matrix by K-means clustering algorithm using the sensor signals, and then, the shortest path algorithm is used to recover source signals, whereas, people suppose that the number of source signals is known when they estimate the mixture matrix by the K-means clustering algorithm generally. In fact, the number of source signals is unknown or blind, so it is very important to estimate the number of source signals, In this paper, a new two-step algorithm is proposed, which not only can estimate the number of source signals but also get the mixture matrix instead of K-means algorithm through the characteristics of sensor signals. The last simulation results show the algorithm simply, efficient and good performance.