智能视频监控系统的声源定位设计与实现  

Design and Implementation of Acoustic Localization in Intelligent Video Surveillance System

作　　者：李磊[1] 李勇[1] 苏永钢[1] 于海锋[1] 

机构地区：[1]郑州大学物理工程学院,河南郑州450001

出　　处：《计算机仿真》2013年第9期378-381,401,共5页Computer Simulation

基　　金：河南省教育厅科学技术研究重点项目(13A140661)

摘　　要：在视频监控系统优化设计的研究中,当前的智能视频监控采用图像处理的跟踪方法。由于摄像头的视角有限,系统存在目标不在视场范围内的监控盲区。为避免上述缺陷,通过声学相控阵给摄像头加上了听觉功能,使监控系统能够自动跟踪声源方位。系统采用麦克风线性阵列接收音频信号,通过端点检测实现有用音频信号的实时检测,通过频域波束形成实现对宽带音频信号的空间定位,最后采用能量值的谱搜索算法定位出声源的方位。前期的仿真和后期基于DSP的嵌入式系统平台实验均验证了改进方法的可行性和工程应用价值。Most conventional cameras can only adopt the method of analyzing the images captured.A blind spot emerges while the targets disappeared due to limited visual angle.For compensating the disadvantage,this paper equipped the camera with auditory function to be capable of tracking the direction of sound source automatically utilizing acoustic phased array.The acoustic signals were received by linear microphone array and judged to be useful or not by endpoint detection.Then frequency domain beamforming algorithm was implemented to acquire spatial orientation of the wideband acoustic signals,whose angle can be calculated by spectrum-search method based on energy value.Both the simulation and late DSP-based embedded system platform have verified the method＇ s feasibility as well as its possibility in engineering application.

关 键 词：视频跟踪 麦克风阵列 波束形成算法 端点检测 

分 类 号：TM930.12[电气工程—电力电子与电力传动]