APP下载

基于倒谱预滤波的二次互相关算法

2021-06-28柳雷雷于玲

电脑知识与技术 2021年10期

柳雷雷 于玲

摘要:传统的广义互相关算法在噪声和混响的情况下效果不佳,而基于倒谱预滤波的时延估计算法尽管改善了广义互相关算法的抗混响效果,但其在噪声环境下性能仍然较差,针对这一缺点,引入二次互相关,同时对相位变换(PHAT)加权函数进行改进,提出了基于倒谱预滤波的二次互相关算法。仿真结果表明:基于倒谱预滤波的二次互相关算法在抗混响和抗噪声的能力上要优于倒谱预滤波时延估计算法和相位变换加权广义互相关(GCC-PHAT)算法,在低信噪比和强混响的条件下仍然可以得到较为准确的时延估计值。

关键词:倒谱预滤波;二次互相关;时延估计;相位变换加权

中图分类号:TN911.7      文献标识码:A

文章编号:1009-3044(2021)10-0008-04

Abstract: The traditional generalized cross-correlation algorithm does not work well in the case of noise and reverberation. The time delay estimation algorithm based on cepstral pre-filtering improves the anti-reverberation effect of the generalized cross-correlation algorithm. However, its performance is still poor in noisy environment. In view of this shortcoming, introduce the second cross-correlation. At the same time, the phase transformation (PHAT) weighting function is improved. A second cross-correlation algorithm based on cepstral pre-filtering is proposed. Simulation results show that: the second cross-correlation algorithm based on cepstral pre-filtering is superior to the delay estimation algorithm based on cepstral pre-filtering and the phase transformation weighted generalized cross-correlation (GCC-PHAT) algorithm in anti-reverberation and anti-noise ability. Under the condition of noise and strong reverberation, more accurate time delay estimation can still be obtained.

Key words: cepstral pre-filtering; second cross-correlation; time delay estimation; phase transformation weighting

1 引言

信号的时延估计是信号处理中经常分析的一个课题,该课题在管道泄漏位置的识别[1-2]、探地雷达[3-4]、声纳工程[5]、微地震监测[6]等方面具有重要意义。在基于时延估计的声源定位应用中,它不仅可以快速定位声源,而且具有较小的计算量,是近年来国内外学者研究的热点。

基于时延估计的声源定位算法是在估计声音信号到达麦克风阵列上各个麦克风间的时间差的基础上,再结合麦克风阵列的空间阵列结构来判断声源的位置,所以准确的估计出时延对基于时延估计的声源定位来说极其重要。目前声源定位中常用的时延估计算法有:广义互相关法(Generalized Cross Correlation ,GCC)、基于LMS(Least Mean Square)的自適应时延估计算法等。传统的广义互相关时延……

登录APP查看全文