【作者】 刘鹏；

【导师】 仪垂杰；

【作者基本信息】 青岛理工大学 ， 信息与通信工程， 2015， 硕士

【摘要】 压缩感知问题已经成为世界各国专家学者讨论研究的热点问题,相比之下运动声源的定位问题出现较早,而研究方向各有不同。声信号和图像信息的处理是实现对运动声源识别的先决条件;与此同时,压缩感知理论在声信号和图像信息处理领域翻开了新的篇章,打破了奈奎斯特定律在信号提取、收集等方法的限制。基于压缩感知理论的运动声源识别技术简化了声源信息的获取以及处理步骤;减少了在获取信号时的硬件操作流程和后台对所获取信号的处理时间。因此,基于压缩感知理论对运动声源识别方法进行研究具有重要意义。(1)基于压缩感知理论的基本原理,通过对压缩感知的三个关键问题即信号的稀疏处理、测量采样和重构进行比较分析,根据运动声源的特性给出了基于离散傅里叶变换(DFT)时变信号稀疏化和特征基匹配追踪方法。(2)分析了运动声源识别的原理,探讨了波束成形识别方法,对运动声源特性做了详细分析,对常见的重构算法做了比较,得到了适用于运动声场的压缩和重建方法即ROMP重构算法。(3)提出了运动声场重建的实验室测试方法,介绍了实验过程中所用到的相关实验设备,分析了对相关的系统组成,检验了实验过程中用到的麦克风阵列性,并对实验环境做了检测,得出了最适合本实验的阵列结构、温度等因素的设置参数。(4)以三种不同的实验方法即声像仪声源识别、LMS阵列识别以及用LMS得到的数据进行压缩重构识别,通过数据及理论分析,得出相关结论:通过压缩重构可以实现运动声源的识别。更多还原

【Abstract】 The problem of compressed sensing has become a hot issue in the world.Compared with the earlier stage,the problem of moving sound source is different.The processing of acoustic signal and image information is the precondition of the realization of the moving sound source identification.At the same time,the compressed sensing theory has opened a new chapter in the field of acoustic signal and image information processing,breaking the limitation of Nyquist’s law in signal extraction and collection.The moving sound source identification technology based on the theory of compressed sensing can simplify the acquisition and processing of the sound source information,and reduce the processing time of the hardware operation flow and the background of the acquired signal.Therefore,based on the theory of compressed sensing,it is very important to study the method of sound source identification.(1)Based on the basic principles of the theory of compressed sensing,by thinning treatments on three key issues,namely compressed sensing signal sampling and reconstruction measuring comparative analysis,based on characteristics of moving sound source is given based on the discrete Fourier transform(DFT)and time-varying signal sparse feature matching based tracking methods.(2)Analyzed the movement principles of sound source identification,discusses beamforming identification method for moving sound source characteristics to do a detailed analysis of the common reconstruction algorithm were compared,has been compressed for sports sound field and reconstruction That ROMP reconstruction algorithm.(3)A laboratory test methods for moving sound field reconstruction,introduced the related laboratory equipment used in the experiment are analyzed related system components,testing the experimental procedure used in an array of microphones,and the experimental environment do the testing,set the parameters obtained in this experiment the most suitable array structure,temperature and other factors.(4)in three different experimental methods namely pan instrument sound source identification,LMS Pattern Recognition and compress data for identification LMSreconstruction obtained through data and theoretical analysis,draw conclusions: the reconstruction can be achieved by compression identify the moving sound source.更多还原