Minimum entropy-based acoustic source localization with Laplace Distribution

Accurate and fast localization of acoustic sources is a problem that is of significant interest in applications such as conference systems, gunshot localization systems. Recently, approaches that based on the time difference of arrival(TDOA) are becoming popular, despite the computational expenses. The TDOA localization algorithms contain two parts: the time delay estimation(TDE) algorithms and space search algorithms. The most important TDE algorithms are based on the generalized cross-correlation(GCC) method. These algorithms perform reasonably well when reverberation or noise is not too high. Lots of improved algorithms are proposed, such as SRP-PHAT. In this article, we show that acoustic source localization with Laplace Distribution can be developed on a basis of minimum entropy (ME). For traditional grid search is too expensive for a real-time system, we propose using stochastic region contraction(SRC) to make computing the ME practical. Results from simulation experiences show excellent accuracy of the ME-SRC algorithm.