去除交叉定位中虚假定位的一种快速算法

测向交叉定位是一种通过对输入源信号进行分析以获得声源定位信息的方法，该方法简便易行，具有较好的应用前景，但是，该方法在多声源的情况下存在虚假定位问题，鉴于这种情况，本文提出一种借助于盲分离来去除假定位的方法，该方法首先通过适应分离算法对源信号进行分离，然后通过实际混迭矩阵的估计和候选混迭矩阵之间的匹配运算以找到正确的定位关系。仿真实验验证了方法的有效性。

A Method for Eliminating False Location in Cross-localization

The cross-localization based on direction measurement is one method to get the location information of the sound sources. The method has some advantages, but there is a problem of false localization in the case of multiple sources. In this paper, a new method of eliminating the false localization based on blind separation is presented. The results of simulation show that the method is an effective method.