| Sound-source locating based on a single array of microphones has been showed many limitations to improve location accuracy and detection area.In recent years,technical researches have hence focused on the system in combining multiple and distributed arrays of microphones.One of fundamental functions of the combination is the coordinate transformation between individual arrays and automatic calibration for practical employment.This thesis provides the design of a system prototype of microphone-array based sound-source locator,and an improved method for position mapping in calibrations.Contributions are as follows.Firstly,the structure of microphone array is analyzed in general in the view of array construction,the applied sound field environment and the signal processing procedure.The implementation approach of the acoustic camera mapping model is proposed based on the beamforming technique in sound-source locating.Secondly,a sound-source locator prototype has been developed based on open-source software,and a driver of a 32-channel sound card and an extension module for real-time video superposition have been implemented to trace the sound source in a visible monitor.Thirdly,for the mutual calibration of multiple and distributed microphone arrays,the epipolar geometry and normalized eight-point algorithm are employed to design calibration computing modules.A numerical simulation verification has also been carried out.Finally,aiming at reducing calibration errors due to the deviation of microphone element in an array,a precise mapping method is proposed to joint self-calibration and iterative mutual-calibration,and the calibration performance is analyzed by numerical simulations.The results show that in comparison with existing methods,the proposed one demands less testing sources for calibration by about 43% in amount,reduces the calibration error about 18% in maximum,and improves the locating accuracy of multiple and distributed microphone arrays. |
PDF Full Text Request