Sound Source Localization Algorithm Based On Compressive Sensing With Microphone Array

Microphone array sound source localization technology development in the late last century,more than at the beginning of this technology used in the war and industry,etc.As the AI,5g,such as the progress of science and technology,People's life is closely combine with the use of intelligent machines is more and more,the microphone array has been widely used in smart home,robot,video conferencing products.This technology has become a part indispensable to these areas,is the most important part of the algorithm to its,so to research the sound source localization algorithm is valuable.In this paper,in view of the traditional algorithm for low frequency signal positioning ability is poor,low resolution of problems,the introduction of compressed sensing theory,compressed sensing algorithm is presented based on the reestablish signal energy.This article main work content is summarized as follows:(1)Described based on microphone array sound source localization algorithm study of specific meaning,lists were reviewed the research history and status quo at home and abroad of all kinds of algorithms.Introduction of compressed perception,Microphone array receiving signals was deduced several different space sparse representation methods for the model,and combined with compression perception theory and through the theoretical estimate the sound source location.Compressed sensing theory in orthogonal matching pursuit algorithm in common use,the classic sound source localization method based on the algorithm for the simulation analysis,at the same time with the traditional algorithm in signal frequency resolution and positioning parameters were compared.(2)Orthogonal matching pursuit algorithm is iterative speed,can achieve the advantages of the perfect solution,but it needs to know the sound source number,namely sound source sparse degree;Or need to set a larger number of iterations,but there is noise when a false reconstruction signals and long computing time,these factors limit the application of the algorithm.Aiming at the problems,put forward by using the method of reconstructing signal energy,energy threshold set in judgment when iteration reconstruction signal whether new source obviously.Make the algorithm in unknown source number under the premise of accurately iteration to the optimal number of signal sparse block sex is introduced at the same time,avoid signal reconstruction caused by the same signal spectrum energy distribution of the error.The simulation verified the algorithm performance,which retain the advantages of orthogonal matching pursuit algorithm at the same time,and good to make up for the shortcomings.(3)Finally using the positioning system is mainly composed of acoustic camera microphone array to verify this improved algorithm,and set up based on the Lab VIEW sound source localization algorithm procedures.To verify the performance of the algorithm in the practical application environment,unused quieter room and other special scenarios,choose daily life square and indoor outdoor office two scenarios were tested respectively.Test set two sound source,the test results show that the improved algorithm in the case of unknown source number can pinpoint the source location,and has better than the traditional algorithm,and simulation test.Statistics improved algorithm each scenario locate at the same time the number of iterations,the results show that the number and setting the sound source maximum error is 0.31.