基于音视频信息融合的目标检测与跟踪算法

针对单一视觉跟踪算法易受遮挡影响的缺陷,提出一种基于音视频信息融合的目标检测与跟踪算法。整个算法框架包括视频检测与跟踪、声源定位、音视频信息融合跟踪3个模块。视频检测与跟踪模块采用YOLOv5m算法作为视觉检测的框架,使用无迹卡尔曼滤波和匈牙利算法实现多目标的跟踪与匹配;声源定位模块采用十字型麦克风阵列获取音频信息,结合各麦克风接收信号的时延计算声源方位;音视频信息融合跟踪模块构建音视频似然函数和音视频重要性采样函数,采用重要性粒子滤波作为音视频融合跟踪的算法,实现对目标的跟踪。在室内复杂环境下对算法性能进行测试,结果表明该算法跟踪准确率达到90.68%,相较于单一模态算法具有更好的性能。     

导向相应功率定位中时延不确定项的影响

在分布式传声器网络应用中,可控波束形成器导向相应功率(SRP)声源定位算法受到导向时延不准确的影响导致定位不稳定.将导向时延不确定项引入SRP定位模型,并对引起不确定项的主要干扰因素进行了分析建模,同时提出了应对导向时延不确定项的SRP定位算法.蒙特卡罗仿真和分析表明该算法具有抑制噪声和降低定位误差的效果,实际实验数据...     

分布式压缩感知麦克风阵列多声源方位估计*

麦克风阵列已被广泛应用于音/视频会议等人机交互领域中时,多声源应用场景对声源方位估计性能提出了更高的要求。压缩感知(CS)声源定位算法将声源定位问题转化为信号的稀疏重构问题,相比传统的定位算法如相位变换加权（SRP-PHAT）和时延累加定位（DS）能够获得较高的定位性能,但多声源的存在一定程度上降低了稀疏程度,影响了CS重构性能。考虑到传统的CS定位算法并未利用多个连续语音帧之间声源空间向量的共同稀疏性,提出采用分布式压缩感知（DCS）理论以改善多声源的稀疏恢复估计的性能。仿真和实验结果表明,相比于传统定位算法和CS-OMP算法,DCS-SOMP算法在不同信噪比和不同声源强度的环境中,对多声源的方位估计都具有更好的定位性能和定位稳健性。     

半经验关系与匹配场联合处理的爆炸声源快速定位

爆炸声源位置的快速准确获取对声源级测量和声传播计算具有重要意义。为了解决利用单一水听器进行爆炸声源定位时难以获得较好的定位效率和精度的问题,提出了一种基于半经验关系与匹配场联合处理的爆炸声源快速定位方法。首先通过爆炸声源满足的半经验关系,对爆炸位置进行预估,缩小匹配参数的搜索范围;同时,在基于多途时延差匹配定位理论的基础上,利用爆炸声源的半经验关系建立联合匹配定位方法,引入气泡脉动周期和冲击波峰值增加匹配物理信息,实现爆炸声源深度和距离精确反演。仿真分析与2013年南海水下爆炸声试验数据分析结果表明,一次气泡脉动周期与多途时延差的联合匹配可提高对爆炸声源深度的估计精度;冲击波峰值与多途时延差的联合匹配可提高对距离的估计精度。额外匹配量的引入减少了估计精度对接收阵元个数的依赖,能够实现用单阵元快速准确地进行爆炸源位置的估计。      

吸顶式传声器阵列阵元坐标标定方法

提出了一种吸顶式传声器阵列阵元坐标的标定方法。针对在混响声场中,时延估计算法性能严重下降从而导致在标定传声器阵元坐标时产生较大误差的问题,提出了利用脉冲声源作为标定声源,并且截取脉冲源直达声的方法来抑制混响声场的影响,提高传声器阵元坐标标定的精度。建立了阵元坐标标定的误差分析模型,并以白噪声和脉冲声源作为标定声源进行数据仿真和对比分析。仿真结果表明,使用脉冲声源作为标定声源能有效地抑制混响声场的影响,获得传声器阵列阵元的准确坐标。同时,在封闭的房间内建立起孔径为3.5 m、64阵元的螺旋状吸顶传声器阵列进行了实验研究,实验结果验证了本文提出方法的有效性。     

吸顶式传声器阵列阵元坐标标定方法

提出了一种吸顶式传声器阵列阵元坐标的标定方法。针对在混响声场中,时延估计算法性能严重下降从而导致在标定传声器阵元坐标时产生较大误差的问题,提出了利用脉冲声源作为标定声源,并且截取脉冲源直达声的方法来抑制混响声场的影响,提高传声器阵元坐标标定的精度。建立了阵元坐标标定的误差分析模型,并以白噪声和脉冲声源作为标定声源进行数据仿真和对比分析。仿真结果表明,使用脉冲声源作为标定声源能有效地抑制混响声场的影响,获得传声器阵列阵元的准确坐标。同时,在封闭的房间内建立起孔径为3.5 m、64阵元的螺旋状吸顶传声器阵列进行了实验研究,实验结果验证了本文提出方法的有效性。      

新型声学超材料单通道麦克风可实现多声源实时定位与分离

在人工系统中,科研人员通常借助由多个传声器组成的传声器阵列来解决声源定位和分离问题。具有高精度声源定位和分离能力的传声器阵列往往需要较大的阵元数量和物理尺寸,这种阵列系统不仅不便于安装和操控,处理多通道信号的计算成本往往也很大,从而导致其应用受限。受生物听音机制的启发,中国科学院噪声与振动重点实验室的博士生孙雪聪与其导师杨军研究员、贾晗研究员等提出了一种基于声学超材料的单通道多声源的定位与分离系统,用一个带有超材料外壳的单通道麦克风实现了三维空间中多个同时发声声源的实时定位与分离。     

二维板声发射源定位实验的一种简易实现方法

对声发射技术中的二维平面定位问题提出一种新的简易实验方案.采用新三角时差算法实现二维板的声源定位和GPS模拟,同时利用简易实验方法,仅用两个超声波传感器完成实验,简化实验器材,降低实验成本,适合普及,计算过程简单,同时也利于物理问题深入地理解.     

运动声源快速定位的声达时差法

针对声达时差法只能用于非运动声源定位的问题,本文提出一种运动声源快速定位方法。该方法以声达时差为基本定位原理,基于声源计算位置对多普勒效应进行解耦并进行声信号多普勒效应修正,根据三角定位方法构建声传播空间矩阵,以声源位置偏差度为目标基于单纯形优化搜索算法进行声源位置快速逼近,实现了对匀速直线运动的单声源的定位追踪,提高定位实时性。该方法将声达时差法拓展到运动声源的定位,同时解决了消除多普勒效应带来的计算过程复杂、运算量大的问题,仅用4个传声器就可实现运动声源的快速定位,突破了传统运动声源识别中对大传声器阵列的依赖。仿真实验和实车运动声源识别实验结果证明了该方法的有效性,本研究为短时发声运动声源的识别提供了一种简便、高效的方法。      

基于粒子滤波的多普勒信息辅助目标定位跟踪算法

多基地声呐探测系统主要通过测量回波的时延和方位信息进行目标定位与跟踪,定位精度受声速、时延和方位测量误差的影响较大,可以通过多普勒信息辅助进一步提高定位跟踪精度。现有的多普勒信息辅助定位跟踪算法多适用于单基地声呐系统,多基地中的多普勒测量值与目标状态的关系更为复杂,需要研究新的融合方法。该文提出了一种适用于多基地声呐系统的多普勒信息辅助采样重要性重采样目标定位跟踪算法,将多普勒信息融入到粒子滤波的重采样过程,使重采样后的粒子集合更逼近目标的真实状态分布,从而提高了目标定位跟踪精度。数值仿真实验结果表明,提出的目标定位跟踪算法可以有效融合多普勒信息,提升目标定位跟踪精度。     

基于声全息方法和双目视觉实现运动声源声场可视化

运动声源声场的可视化是一种重要的运动声源定位的技术手段,利用双目视觉测量技术实现运动声源声场空间的自动测量,自动确定运动声源表面的空间位置,针对声源表面,利用传声器阵列,基于声全息方法实现运动声源声场的重建,建立视频图像与声场的空间映射,并建立视频与声场之间的时序,实现实景视频图像与声场重建结果的融合,可以自动生成声源运动过程的视频。基于该方法所开发了一套试验测量系统,对运动声源的测量试验结果表明,该方法可以有效实现运动声源的视频可视化,使人可以直接从视频中看到声源及其变化过程,使声源的定位和识别变得更加简单。      

一种利用分布式传声器阵列的声源三维定位方法

为了提高噪声和混响条件下分布式传声器阵列进行声源定位的性能,提出一种利用空间稀疏性和压缩感知原理的声源三维定位方法。该方法首先通过两次离散余弦变换方式提取出声音信号特征,并用该特征来构建稀疏定位模型,以便能够综合利用语音信号的短时和长时特性,同时降低模型维数;然后利用在线字典学习技术动态调整字典,克服稀疏模型与实际信号之间的失配问题,增强稀疏定位模型的鲁棒性;进而提出一种改进的平滑l₀范数稀疏重构算法来进行声源位置解算,以提高低信噪比条件下的重构精度。仿真结果表明该方法不仅可以实现多目标定位,而且具有较强的抗噪声和抗混响能力.      

Video visualization for moving sound sources based on binoculars stereo and acoustic holography

The video visualization methods for moving sound sources are very important. The spatial positions in the sound filed measurement of moving sources arc measured by an automated method based on the binoculars stereo technique,the spatial coordinates of the surface of the moving sound source are determined automatically.The sound field of the surface of the moving source is reconstructed with a microphone array using the acoustic holography method for moving sound sources.The spatial mapping between the sound field and the video images is constructed,and also the time sequence of the sound field and the images is founded. Then,the reconstructed sound field and the real-world images are merged,and the visualized sound field video of the moving source is generated automatically.A sound field measurement system based on this method is developed,the measurement results of the moving sound source with this system indicate that the sound field video visualization is realized effectively with the proposed method,the sound sources and their changing process can be seen directly from the generated video,the identification and localization for the moving sound sources can be easier with this method.     

Robust tracking of moving sound source using multiple model Kalman filter

In this paper a novel method for tracking an active speaker in a noisy and reverberant environment by means of a spatially distributed microphone array is presented. Firstly, a sound source localization algorithm based on time delays of arrival (TDOA) in microphone pairs provides observed position estimates. Then these remarkably noisy estimates are filtered by a multiple model Kalman filter (MMKF) in order to obtain a smoothed trajectory of the speaker’s movement. Compared with the traditional Kalman filter (KF), simulated results prove the MMKF is more robust and effective in noisy environments.     

Three-dimensional sound source localization using distributed microphone arrays

To improve the performance of sound source localization based on distributed microphone arrays in noisy and reverberant environments,a sound source localization method was proposed.This method exploited the inherent spatial sparsity to convert the localization problem into a sparse recovery problem based on the compressive sensing(CS) theory.In this method two-step discrete cosine transform(DCT)-based feature extraction was utilized to cover both short-time and long-time properties of the signal and reduce the dimensions of the sparse model.Moreover,an online dictionary learning(DL) method was used to dynamically adjust the dictionary for matching the changes of audio signals,and then the sparse solution could better represent location estimations.In addition,we proposed an improved approximate l_0norm minimization algorithm to enhance reconstruction performance for sparse signals in low signal-noise ratio(SNR).The effectiveness of the proposed scheme is demonstrated by simulation results where the locations of multiple sources can be obtained in the noisy and reverberant conditions.     

三维空间非同步测量高分辨率声成像方法

为解决三维空间中声源成像分辨率低的问题,提出一种基于非同步测量的三维空间声成像方法.该方法首先通过移动球形传声器阵列扫描空间分布的声源,然后利用非同步测量技术近似得到大孔径、高密度的传声器阵列测量结果,最后通过传统波束形成算法成像.仿真及实验结果表明,该方法与单次测量下的波束形成方法相比,聚焦性能更好,空间分辨率高,可...     

Blind extraction and localization of sound sources using point sources based approaches

This paper presents theoretical models for blind sound source localization and separation of the signals emitted by arbitrary point sources in free space. Source localizations are achieved by a model based approach that accounts for the spherical spreading of an acoustic wave and utilizes an iterative triangulation, based on the signals measured by a three-dimensional microphone array. Once source locations are determined, the source signals are separated by using the point source separation (PSS) method, which is valid for all types of signals, including harmonic, continuous, transient, random, narrowband and broadband. General solutions for signals separation are presented. Theoretically, PSS can reconstruct the individual source signals exactly. This is because it employs the free-space Green's function, which defines the exact correlation among individual sources and measurement microphones. To validate PSS, numerical simulations are carried out and results are compared with those obtained by FastICA (Independent Component Analysis) code. The impacts of various parameters such as the microphone configuration, type of source signals, signal to noise ratio, number of microphones and source localization errors on the quality of signals separation by using PSS and FastICA are examined. The advantages and disadvantages of PSS and FastICA are compared and discussed.     

基于蒙特卡罗法的实时空间三维声源定位

构建了一个基于四个声音传感器的信号时延采集系统,根据采集系统得到的三个时间差和传感器的响应顺序,提出了一种基于蒙特卡罗法实时空间的三维声源定位算法.该算法通过三维声音定位的非线性方程,构建一个三维模函数,通过寻找空间全局收敛点,并根据公差容限进行变步长搜索,准确快速地计算出声源的位置.     

Localisation of moving sources in a multipath propagation environment

A new algorithm is presented for tracking correlated narrow-band sources in the presence of colored Gaussian noise. A fast cumulant-based preprocessing method is used to remove unknown noise and a Kalman filtering is used to track the source parameters. The use of a Kalman filtering avoids the data association problem and improves the tracking performance for crossing tracks. It is applied to the outputs of Newton’s algorithm to track moving sources. In this paper, the algorithm is developed for the special case in which the updated cumulant matrix is obtained by substituting a new matrix of the current data. The rank tracking problem is not considered in this study.We demonstrate the performance of the proposed algorithm by computer simulations of the tracking of moving targets emitting correlated signals, we also tested the proposed algorithm on the real data recorded during an underwater acoustic experiments.