基于时延矢量封闭准则的多传声器降噪方法

目前国内外提出的经验模式分解降噪方法大都在已知噪声类型的基础上针对单传声器信号进行研究,没有利用多传声器之间的融合信息。因此,提出基于时延矢量封闭准则的多传声器信号降噪方法。该方法首先将采集得到的信号进行EMD分解,将所得的对应IMF分量进行互相关,求取时延值,然后依据TDVCR进行时延矢量误差运算,得到IMF分量权重矩阵,最后根据IMFs及其对应的权重矩阵进行信号重构,得到去噪后的信号。试验表明:该方法有效利用了多传声器时延矢量匹配准则,在不增加任何先验信息的前提下降低了噪声,取得了满意的试验效果。

A Multi-microphone Denoising Algorithm Based on Time Delay Vector Close Rule

The present denoising method with EMD(empirical mode decomposition) usually needs to know the type,and it is only used for one sensor,which ignores information fusion of multi-sensor.A new denoising algorithm with information fusion of multi-sensor based on TDVCR(time delay vector close rule) was proposed in this paper.Firstly,empirical mode decompositions of the collected original signals were carried out,and time delays were computed through the correlation of corresponding IMF components.Then,time delay ...