中文摘要：

针对空间声源水平方位定位的精度问题,提出了利用基于软域值的小波分析方法对声源信号进行去噪处理.在此基础上,利用机器人的听觉系统对目标声源的水平方位进行粗定位;通过双目立体视觉系统对粗方位进行一定的矫正、补偿,实现方位的精定位.实验证明,提出的由粗到精的定位策略和方法具有较高的定位精度.

英文摘要：

For the azimuth accuracy of sound source localization, this paper adopted wavelet analysis to denoise signal with soft-threshold. Based on this, estimated the rough azimuth degree using the auditory system of robot. In order to gain the fine direction of azimuth, paper provided the error compensation model of azimuth with binocular stereo vision system. Through experiments, the provided coarse-to-fine strategy and algorithm can generates more accurate results for sound source localization.