基于检测前跟踪的声源跟踪算法

Acoustic source tracking algorithm using track before detect

针对低信噪比、强混响环境中声源跟踪误差较大的问题,提出一种基于检测前跟踪的声源跟踪算法。该算法将一种改进的可控响应功率函数作为定位函数,通过计算粒子所在矩形区域的可控响应功率值,改进的可控响应功率函数可以得到更加顽健的声源位置估计;同时,将检测前跟踪技术应用于跟踪算法中,可以免于重复计算处于同一区域的量测值,从而在保证跟踪精度的前提下大幅减少跟踪算法的运算量。经仿真实验验证,改进算法在低信噪比、强混响的环境中能够得到比传统跟踪算法更精确的跟踪结果。

In order to improve tracking performance in a noisy and reverberant environment,an acoustic source tracking algorithm using track before detect was proposed.This algorithm used a modified steered response power as localization function which take into account a rectangular region to achieve more robust source location estimation than steered response power function and applied track-before-detect technology to avoid reduplicate calculation of the same rectangular region so that the algorithm could reduce the computation burden without decreasing the accuracy.The simulation results verified that the proposed algorithm can achieve more accurate tracking results than traditional tracking algorithm in a noisy and reverberant environment.