【摘要】 针对强干扰环境下的特定声方向角(DOA)定位问题,提出了一种基于多重信号分类(MUSIC)和模式识别的定位方法。强干扰声是指其功率大于特定声的干扰声。首先利用MUSIC算法把特定声和强干扰声的DOA都估计出来,然后结合波束形成和模式识别技术从这几个方向角中挑选出特定声的方向角,从而实现了在强干扰环境下的特定声方向定位。实验证明,干扰声功率是有效声功率的10～100倍时,该方法能以78%以上的准确率估计出特定声的方向角。更多还原

【Abstract】 A system to localize specific sounds in low SNR environments is presented. It consists of three parts: effective estimation of DOAs(Directions of Arrival) of all the sounds with MUSIC(Multiple Signal Classification), recognition of the direction of the specific sound with a beam former and a classifier. The field test shows that the system is able to localize gun sounds in 10～20 dB higher noises, with accuracy no less than 80%.更多还原