Research On Passive Acoustic Localization Based On Multiple Microphone Arrays

Passive acoustic localization based on the microphone array,that has only one receiver,can localize the distance and bearing of a signal by receiving unintentionally or intentionally radiation signal from target.It is the hot study topics in the acoustic source localization because of the strong concealment,low cost,high safety,and so on.Passive localization based on microphone array technology is mainly done in two steps.First,the target acoustic signal is received by the microphone array,and localization equations with TDOA(time delay of arrival)parameter are established based on geometrical structure of microphone array.Second,different coordinates of microphone elements lead to the different propagation distance and TDOA between the acoustic signals at the microphones.The location parameters of acoustic source will be obtained by estimating and substituting TDOAs into the established localization equations.Therefore,geometrical structure of microphone array and TDOA estimation are the t wo important factors influencing the localization accuracy in the passive localization of acoustic source.However,there are some problems such as the low accuracy,large computational complexity and poor real-time because of the sound signal particularity and the sound field complexity.So far,many methods devote to both improve the localization accuracy and reduce the localization computation time.For human-computer interaction and large videoconferencing,reflection waves,cross-correlation noise and sampling frequency will influence on accuracy of TDOA estimation,as well as distance error is bigger based on single microphone array.Therefore,this paper studies the improvement of the localization accuracy,as well as pays attention to the simplicity and real-time performance of the proposed methods.We analyze the different structure of microphone array for collecting a spatial acoustic source.A joint microphone array with multiple sub-array is established based on an optimal microphone array with four microphones single,witch focus on improving localization accuracy and real-time performance.On that basis,data fusion method combining the optimum extremum algorithm is studied to localize the acoustic source.For estimating the TDOA in passive loc alization of an acoustic source,GCC(generalized cross-correlation)method used with the MIUS(multistage implementation of up-sampling)can improve localization accuracy under the conditions of low sampling rate.At the same time,based on Snells law,the proposed virtual microphone array is combined with wavelet threshold denoising method for further filtering processing of the acoustic signal,and further improving the localization accuracy of the acoustic source in complex noise environment.Specific studies include the following aspects.This paper proposes a joint microphone array with multiple sub-microphone arrays based on analyzing the structure of microphone array(i.e.geometric construct,spacing between array elements,and number of microphone s),which has the wider localization range and higher localization accuracy comparing with single microphone array.In order to reduce the localization computation time and data amount,every two sub-microphone arrays will share a microphone element.At the same time,via data fusion based on the new multiple microphone arrays,an optimal objective function including acoustic source location parameters can be established,and then solved using the optimal search method to obtain the target location estimates.Furthermore,approximate localization errors' curves associated with the acoustic source location parameters(namely,distance,azimuth angle and pitch angle)can be obtained via analyzing the localization equations of microphone array.To analyze the relations between sampling rate of acoustic source and localization accuracy based on the GCC method,we proposes a modified GCC method with the MIUS which is associated with increasing the original sampling rate of input signal,for improving the source location accuracy in the low sampling rat.In this paper,an explicit solution to a total interpolation factor and its optimal decomposition scheme is presented through the simulations and mathematical analysis,which relates to both effectively improving the localization accuracy and saving the computation time.By analyzing the propagation model of the acoustic source in the semi-free acoustic field,this paper proposes the virtual microphone array,which is combined with the original real microphone array to form a new microphone array according to Snells law.All of TDOA estimations using GCC method are substituted into the established new localization equations to solve the acoustic source location parameters.Furthermore,the localization performance b y the GCC method will decline when the collected acoustic signals are a non-stationary signal,and noises between collected channels are correlative.Many studies have shown that the GCC method based on the wavelet transform can well improve the localizati on accuracy of the non-stationary signal with the correlated noises.Therefore,the wavelet threshold filter is employed in the GCC method to improve the acoustic localization accuracy.In this work,an improved adaptive wavelet threshold is proposed based on the change rules of the wavelet coefficients to further obtain the better de-noising effects.A set of simple passive acoustic localization system is established,and localization experiments indoor are completed based on this system.The proposed joint microphone array with multiple sub-arrays is applied in receiving the target signal(sampling frequency of 8 KHZ).Then,the collected signal data through multistage implementation of up-sampling process,virtual microphone array structure and the wavelet threshold denoising,are fused to obtain a more accurate location parameters of the acoustic source.Finally,the obtained experimental results are compared with those in other literature to verify the effectiveness of the proposed methods in this paper for improving localization accuracy.