低信噪比混响环境下声测定位方法的实现

发布时间：2018-04-01 23:15

本文选题：声源定位　切入点：数学形态学　出处：《电子科技大学》2014年硕士论文

【摘要】：狭小空间中的低信噪比混响环境会影响声源定位系统的定位精度,严重时甚至会致使系统无法正常工作。即使利用滤波器对噪声进行滤波处理,声音信号也会因被障碍物反射而存在着一定的混响,常规滤波器无法进行滤除。由此可见,提出一种能滤除混响和噪声且定位精度高的方法十分必要。针对上面实际中遇到的问题,本文以四元T型麦克风阵列为例,针对其接收到含有低信噪比噪声和混响的声音信号为研究对象,在深入分析形态滤波器滤除噪声和信号的经验模态分解和重构的基础上,提出了一种适用于低信噪比混响环境的基于数学形态学的声源定位方法,主要内容如下:首先,为解决低信噪比噪声中单独使用小波滤波器时的频率分辨率较低以及对存在跳变信号时处理效果不佳的问题,提出了一种结合形态学滤波器和小波滤波器的融合型滤波器——形态-小波滤波器对缺陷信号进行处理。本文以混有white噪声,destroyerengine噪声和f16噪声的信号为例,针对其信号的特性,运用平结构元素的级联结构的形态学滤波器提高低信噪比信号的信噪比,再由小波滤波器进行滤波处理,实现对低信噪比信号的噪声滤除。其次,在前面滤除噪声过程的基础之上,将小波的预处理方法与经验模态分解方法结合起来,分离混在声音信号中的混响信号,得出原信号的重构信号。本文以一段加上混响后的语音信号为例,采用小波预处理方法将信号分为一系列窄带信号,然后再使用经验模态分解,得到一系列的能表征信号特征的信号分量,筛选后重构原信号,最终就可以得到滤除混响的声音信号。最后,根据前面的滤除噪声和混响的方法并结合基于时延估计的声源定位算法,设计了基于数学形态学和经验模态分解的声测定位方法,并在mini2440上使用Matlab实现了声测定位系统。经过仿真实验可知,该方法可以有效地滤除噪声,抑制混响,为进一步提高定位精度,增强抗外界环境因素的干扰性能奠定了基础。
[Abstract]:The reverberation environment of low signal-to-noise ratio in a narrow space will affect the positioning accuracy of the sound source location system and even cause the system to fail to work properly even if the noise is filtered by the filter. The sound signal will also be reflected by obstacles and there will be reverberation, which can not be filtered by conventional filters. It is necessary to propose a method to filter reverberation and noise with high positioning accuracy. In view of the problems encountered in practice above, this paper takes the quaternion T microphone array as an example. For the acoustic signal with low SNR noise and reverberation received, the empirical mode decomposition and reconstruction of filtering noise and signal by morphological filter are deeply analyzed. A mathematical morphology based sound source location method for low SNR reverberation environment is proposed. The main contents are as follows: firstly, In order to solve the problem of low frequency resolution when wavelet filter is used alone in low signal-to-noise ratio (SNR) noise, and the problem that the processing effect is not good when there is a jump signal, In this paper, a hybrid filter, Morphology-wavelet filter, which combines morphological filter and wavelet filter, is proposed to process the defective signal. In this paper, the signal mixed with white noise, destroyerengine noise and f16 noise is taken as an example, and the characteristics of the signal are analyzed. A cascade morphological filter with flat structure elements is used to improve the signal-to-noise ratio (SNR) of low signal-to-noise ratio (SNR) signal, and then the wavelet filter is used to filter the signal to realize the noise filtering of the low signal-to-noise ratio signal. Secondly, On the basis of filtering noise process in front, the pre-processing method of wavelet and empirical mode decomposition method are combined to separate reverberation signal mixed in sound signal. The reconstructed signal of the original signal is obtained. Taking a speech signal with reverberation as an example, the signal is divided into a series of narrow band signals by wavelet pretreatment method, and then the empirical mode decomposition is used. We get a series of signal components that can represent the characteristics of the signal. After screening the original signal, we can reconstruct the original signal, and finally we can filter the reverberation sound signal. Finally, According to the previous methods of noise filtering and reverberation, combined with the acoustic source location algorithm based on time delay estimation, the acoustic location method based on mathematical morphology and empirical mode decomposition is designed. The sound measurement and positioning system is realized on mini2440 with Matlab. The simulation results show that the method can effectively filter noise and suppress reverberation, which lays a foundation for further improvement of positioning accuracy and enhancement of anti-interference performance of external environment factors.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN912.2

【参考文献】