摘要

For sound source localization in enclosed space, the background noise and reverberation have significant influences on traditional localization algorithms and the array, and these algorithms are usually complex. A sound source localization method called dual-channel matching is proposed in this paper. The method is based on the theory of time reversal. It has strong anti-reverberation capability and the ability to overcome multipath effects. A series of sound source localization experiments were carried out in an underground parking lot, where the real-time performances of the method under different dimensions of sound source devices, SNR, and reverberation were evaluated. The parameters regarding the microphone, obstacle's influence, different heights between sampling and localization processes, and localization of slow-moving sound sources were also evaluated. The results and their analysis show that (1) high spatial resolution can be obtained while the sound devices satisfy the requirement of the radiated acoustic energy, and that the smaller the size, the smaller is the error range which corresponds to higher spatial resolution; (2) when the SNR is greater than 2 dB, spatial resolution in common room is 10 cm; (3) spatial reverberation has little effects on localization resolution; (4) the method has strong tolerance for environment; and (5) when the sound source moves slowly, this method also has a good localization effect.