基于粒子滤波的麦克风阵列声源定位方法  

Particle Filter for Acoustic Source Localization

作　　者：李彦翔[1] 刘庆华[1] 

机构地区：[1]桂林电子科技大学信息与通信学院,桂林541004

出　　处：《微计算机信息》2010年第7期233-235,共3页Control & Automation

基　　金：基金申请人:刘庆华;项目名称:语音定位技术研究;基金颁发部门:广西自然科学基金(0832007Z)

摘　　要：针对噪声与混响环境下的声源定位问题,本文采用了一种基于粒子滤波的麦克风对的声源定位方法。该方法在粒子滤波框架下,将到达麦克风对的时间差作为观测信息,通过计算麦克风对的广义互相关函数(GCCF)来构建似然函数。实验结果表明,本文所采用的方法提高了声源定位系统的抗噪声与抗混响能力,即使在低信噪比强混响的环境下也能获得较高的定位精度。A sound source localization method based on microphone pairs is presented for Acoustic Source Localization problem in noisy and reverberant environments.In the framework of particle filtering,the Time Delay of Arrival（TDOA）is servered as the observation information,and likelihood function is constructed by the generalized cross-correlation（GCC）function.The results of simulation data and real data show that the proposed method can reliably confirm the Acoustic Source＇s position even in moderately reverberant rooms.Even in the environment of low SNR and strong reverberation,the adopted method can also give effectual localization.

关 键 词：混响环境 粒子滤波 到达时间差 声源定位 均方根误差 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]