逆模型下的高效声源定位算法研究

与波束形成算法相比，广义互相关逆模型宽带声源定位算法提供了空间的高分辨率，但需要更高的计算量。为了提升广义互相关逆模型算法的计算效率，并尽可能保留其分辨率优势，本文提出了一种高效的声源定位算法。该算法首先去除麦克风阵列输出功率较小的网格点来压缩计算网格，其次，在几何因素条件下利用基于密度的聚类进一步压缩网格，最后仅用保留下的点进行计算，从而大大降低了计算规模。真实数据实验证明，本文所提算法能有效提升广义互相关逆模型算法的计算效率。

Efficient acoustic source localization algorithm based on an inverse model

Compared with the beamforming algorithm, the wideband acoustic source localization based on the generalized crosscorrelation inverse model provides a higher spatial resolution and requires more computation. This paper proposes an acceleration algorithm to improve the computational efficiency of acoustic source localization and retain the existing resolution. By removing the grid points with the low output power of the microphone array, the computational grid is compressed. The geometric factor is also considered, and the density-based clustering algorithm is used to further compress the grid. Finally, only the reserved points are used for the calculation to reduce the calculation scale. Experiments with real data show that the proposed algorithm can improve computational efficiency significantly.