Noise reduction combining time-frequency epsilon-filter and M-transform

This paper introduces noise reduction combining time-frequency epsilon-filter (TF epsilon-filter) and time-frequency M-transform (TF M-transform). Musical noise is an offensive noise generated due to noise reduction in the time-frequency domain such as spectral subtraction and TF epsilon-filter. It has a deleterious effect on speech recognition. To solve the problem, M-transform is introduced. M-transform is a linear transform based on M-sequence. The method combining the time-domain epsilon-filter (TD epsilon-filter) and time-domain M-transform (TD M-transform) can reduce not only white noise but also impulse noise. Musical noise is isolated in the time-frequency domain, which is similar to impulse noise in the time domain. On these prospects, this paper aims to reduce musical noise by improving M-transform for the time-frequency domain. Noise reduction by using TD M-transform and the TD epsilon-filter is first explained to clarify its features. Then, an improved method applying M-transform to the time-frequency domain, namely TF M-transform, is described. Noise reduction combining the TF epsilon-filter and TF M-transform is also proposed. The proposed method can reduce not only high-level nonstationary noise but also musical noise. Experimental results are also given to demonstrate the performance of the proposed method.     

基于时频分析的多目标盲波束形成算法

针对现有盲波束形成算法适用范围较窄, 多目标信号分离级联模式结构复杂、并联模式稳定性较差等问题, 提出一种基于时频分析的多目标盲波束形成算法. 该算法首先利用时频分析技术给出信号导向矢量的不确定集, 然后优化求解导向矢量的最优估计, 最后利用Capon方法实现多目标信号的并行输出. 理论分析及仿真结果表明, 该算法对信号特性没有特殊要求, 适用性较广, 性能稳定, 且输出信干噪比高于其他盲波束形成算法, 接近于最优Capon波束形成器.     

水声JANUS信号的分数低阶时频谱迁移学习识别方法

非合作第三方水下标准协议信号识别在水声通信信号识别中具有重要研究意义。针对浅海水声JANUS信号的特征提取因易受脉冲噪声和多径效应等复杂水声环境影响而导致识别率低下的问题,提出一种分数低阶时频谱和ResNet18 （Residual Network 18）相结合的迁移学习识别方法。首先,选取JANUS固定前导作为识别对象,设计分数低阶傅里叶同步压缩变换（FLOFSST）,以分数低阶操作抑制脉冲噪声,以时频重排特性增强时频集中性。其次,将基于ImageNet的ResNet18预训练模型微调,迁移至JANUS信号和常见水声信号时频图集。仿真表明所提算法在信噪比为-10 dB时JANUS信号的识别率为96.15%,能够有效抑制脉冲噪声并减小多径效应影响,比传统算法识别性能好。海试中JANUS信号识别率达90.00%,证明算法识别准确率和网络的泛化性较高。      

利用小波多尺度分解算法实现混沌系统的噪声减缩

应用小波多尺度分解算法进行噪声减缩,从混沌背景中分离周期信号、噪声及其他混沌信号.小波多尺度分解算法能够区分不同尺度的信号是利用小波变换在时、频两域具有突出信号特征的能力以及小波变换是一线性变换的特点.提出的方法仅利用信号的尺度特性,克服了先前的噪声减缩要知道产生混沌信号的数学模型,并且要求叠加在混沌背景中的其他信号的幅度相对混沌背景信号的幅度很小的假定.给出了从Lorenz混沌背景中提取正弦信号、白噪声和Chua's电路产生的混沌信号的计算机模拟结果. 关键词：     

联合深度编解码网络和时频掩蔽估计的单通道语音增强

提出了一种联合深度编解码神经网络和时频掩蔽估计的语音增强方法。该方法利用深度编解码网络估计时频掩蔽表示,并联合带噪语音的幅度谱学习带噪语音与纯净语音幅度谱之间的非线性映射关系。深度编解码网络采用卷积-反卷积网络结构。在编码端,利用卷积网络的局部感知特性,对带噪语音的时频域结构特征进行建模,提取语音特征,同时抑制背景噪声。在解码端,利用编码端提取到的语音特征逐层恢复局部细节信息并重构语音信号。同时,在编解码端对应层之间引入跳跃连接,以减少由于池化和全连接操作导致的低层细节信息丢失的问题。在TIMIT语音库和不完全匹配噪声集下进行仿真实验,实验结果表明,该方法可以有效抑制噪声,且能较好地恢复出语音细节成分。      

混沌干扰中基于同步挤压小波变换的谐波信号提取方法

针对混沌干扰背景下多个谐波信号的提取问题, 提出了一种基于同步挤压小波变换(SST)的谐波信号抽取方法. 首先利用SST将混沌信号和谐波信号组成的混合信号分解为不同的内蕴模态类函数, 然后利用Hilbert变换对分离出的内蕴模态类函数进行频率识别, 从中分离出各谐波信号. 以Duffing混沌背景为例, 对混沌干扰下多谐波信号的提取进行了实验分析. 实验结果表明: 对于不同频率间隔的多个谐波分量, 本文方法的提取结果都具有较高的精度, 而且所提方法对高斯白噪声的干扰具有较好的鲁棒性, 综合提取效果优于经典的经验模态分解方法.     

结合振动控制的柱面纵向梯度线圈目标场设计方法

在磁共振成像系统的工作过程中,噪声主要是由梯度线圈系统产生的.梯度线圈置于高均匀度超导磁体的室温孔内,并工作于脉冲状态,频繁的开启和关闭会使线圈中电流急剧随时间变化,变化的电流导致线圈受到变化的洛伦兹力作用,从而产生振动,这种高频振动所发出的噪声会对病人产生刺激,严重时甚至会对病人的听觉神经产生损伤.梯度场的场强越强、切换速度越快,所产生的噪声就越大.降低噪声的最根本方法是通过有效的梯度线圈设计,降低洛伦兹力的空间分布.本文针对纵向梯度线圈,在原经典目标场设计方法基础上,加入对振动参量,从而能够有效地降低线圈工作时所产生的噪声.其具体方法是将振动控制函数作为约束条件,通过目标场法建立数学模型,利用MATLAB进行电磁验算.计算结果表明,所提数学模型可有效地降低线圈振动的最大振幅.     

心磁信号广义S变换域奇异值分解滤波方法

针对心磁信号工频及背景噪声干扰问题, 提出了广义S变换奇异值分解(singular value decomposition, SVD)滤波方法.在离散S变换基础上, 导出了广义矩阵S变换和逆变换公式. 通过对采样信号进行广义S变换, 调节时频分辨率, 利用SVD分解方法确定有效心磁信 号区域, 实现自适应时频滤波. 实验结果表明, 该方法能有效滤除工频及背景噪声干 扰, 且在较少奇异值个数情况下可获得更好的滤波性能. 关键词： 心磁信号 S变换 奇异值分解 时频滤波     

Spectrogram denoising and automated extraction of the fundamental frequency variation of dolphin whistles

Marine mammal vocalizations are often analyzed using time-frequency representations (TFRs) which highlight their nonstationarities. One commonly used TFR is the spectrogram. The characteristic spectrogram time-frequency (TF) contours of marine mammal vocalizations play a significant role in whistle classification and individual or group identification. A major hurdle in the robust automated extraction of TF contours from spectrograms is underwater noise. An image-based algorithm has been developed for denoising and extraction of TF contours from noisy underwater recordings. An objective procedure for measuring the accuracy of extracted spectrogram contours is also proposed. This method is shown to perform well when dealing with the challenging problem of denoising broadband transients commonly encountered in warm shallow waters inhabited by snapping shrimp. Furthermore, it would also be useful with other types of broadband transient noise.     

Optimum adaptive array stochastic resonance in noisy grayscale image restoration

Considering the widespread noise interference in the two-dimensional (2D) image transmission processing, we proposed an optimal adaptive bistable array stochastic resonance (SR)-based grayscale image restoration enhancement method under low peak signal-to-noise ratio (PSNR) environments. In this method, the Hilbert scanning is adopted to reduce the dimension of the original grayscale image. The 2D image signal is converted into a one-dimensional (1D) binary pulse amplitude modulation (BPAM) signal. Meanwhile, we use the adaptive bistable array SR module to enhance the 1D low SNR BPAM signal. In order to obtain the restored image, we transform the enhanced BPAM signal into a 2D grayscale image signal. Simulation results show that the proposed method significantly outperforms the classical image restoration methods (i.e., mean filter, Wiener filter and median filter) both on the grayscale level and the PSNR of the restored image, particularly in a low PSNR scenario. Larger array size brings better image restoration effect.     

Fine structure spectrography and its application in speech

A filterbank-based algorithm for time-varying spectral analysis is proposed. The algorithm, which is an enhanced realization of the conventional spectrogram, consists of hundreds or thousands of highly overlapping wideband filter/detector stages, followed by a peak detector that probes the filter/detector outputs at very short time intervals. Analysis with synthetic modulated signals illustrates how the proposed method demodulates these signals. The resulting spectrogram-like display, referred to as a "fine structure spectrogram," shows the fine structure of the modulations in substantially higher detail than is possible with conventional spectrograms. Error evaluation is performed as a function of various parameters of a single- and two-component synthetic modulated signal, and of parameters of the analysis system. In speech, the fine structure spectrogram can detect small frequency and amplitude modulations in the formants. It also appears to identify additional significant time-frequency components in speech that are not detected by other methods, making it potentially useful in speech processing applications.     

Wavelet filter-based weak signature detection method and its application on rolling element bearing prognostics

De-noising and extraction of the weak signature are crucial to fault prognostics in which case features are often very weak and masked by noise. The wavelet transform has been widely used in signal de-noising due to its extraordinary time-frequency representation capability. In this paper, the performance of wavelet decomposition-based de-noising and wavelet filter-based de-noising methods are compared based on signals from mechanical defects. The comparison result reveals that wavelet filter is more suitable and reliable to detect a weak signature of mechanical impulse-like defect signals, whereas the wavelet decomposition de-noising method can achieve satisfactory results on smooth signal detection. In order to select optimal parameters for the wavelet filter, a two-step optimization process is proposed. Minimal Shannon entropy is used to optimize the Morlet wavelet shape factor. A periodicity detection method based on singular value decomposition (SVD) is used to choose the appropriate scale for the wavelet transform. The signal de-noising results from both simulated signals and experimental data are presented and both support the proposed method.     

Adaptive autoregressive modeling of non-stationary vibration signals under distinct gear states. Part 2: experimental analysis

Parametric time-frequency representation based on parametric models is more desirable for presenting highly precise time-frequency domain information due to its high-resolution property. However, the sensitivity and robustness of parametric models, in particular the parametric models on the basis of advanced adaptive filtering algorithms, has never been investigated for on-line condition monitoring of rotating machinery. Part 1 of this study proposed three adaptive parametric models based on three advanced adaptive filtering algorithms. Part 2 of this study is concerned with the effectiveness of the proposed models under distinct gear states, especially the highly non-stationary conditions accrued from advanced gear faults. Four gear states are considered: healthy state, adjacent gear tooth failure, non-adjacent gear tooth failure and distributed gear tooth failure. The vibration signals used in this study include the time-domain synchronous averaging signal and gear motion residual signal for each considered gear state. The test results demonstrate that the optimum filter behavior can readily be attained and the white Gaussian assumption of innovations can relatively be easily guaranteed for the NAKF-based model under distinct gear states and a wide variety of model initializations. On the other hand, the EKF- and MEKF-based models are capable of generating more accurate time-frequency representations than the NAKF-based model, but in general the optimality condition for white Gaussian assumption cannot be guaranteed for these two advanced models. Therefore, the NAKF-based model is preferred for automatic condition monitoring due to its appealing robustness to distinct gear states and arbitrary model initializations, whereas the EKF- and MEKF-based models are desirable when accurate time-frequency representation is concerned.     

数字滤波在心磁噪声抑制中的应用

提出了一种基于数据重置最小均方算法的自适应数字滤波器, 并将其应用于抑制心磁信号的背景噪声. 数值模拟结果表明该滤波器对于相关噪声,尤其是有幅值变化和时间延迟的情况, 有很强的抑制效果. 该滤波器有很宽的带宽.利用它可以使心磁信号的信噪比提高到器件的本征噪声水平. 利用周期平均的方法进一步提高信噪比, 对滤波前后心磁信号的频谱进行了分析.     

基于字典学习和稀疏表示的单通道语音增强算法综述*

如何从带噪语音信号中恢复出干净的语音信号一直都是信号处理领域的热点问题。近年来研究者相继提出了一些基于字典学习和稀疏表示的单通道语音增强算法,这些算法利用语音信号在时频域上的稀疏特性,通过学习训练数据样本的结构特征和规律来构造相应的字典,再对带噪语音信号进行投影以估计出干净语音信号。针对训练样本与测试数据不匹配的情况,有监督类的非负矩阵分解方法与基于统计模型的传统语音增强方法相结合,在增强阶段对语音字典和噪声字典进行更新,从而估计出干净语音信号。本文首先介绍了单通道情况下语音增强的信号模型,然后对4种典型的增强方法进行了阐述,最后对未来可能的研究热点进行了展望。     

基于平均吸收的太赫兹波振幅成像研究

提出了一种基于一定频率内平均吸收的太赫兹(THz)波振幅成像新方法。太赫兹波频率在0.1～10 THz之间,波段位于红外和微波之间。太赫兹波成像技术的一个显著特点是信息量大,如何对每个样品点的大量信息进行处理提取有用信息重构出样品的图像是一项关键技术。选用中间挖空有“THz”字样的白纸为样品作太赫兹波成像研究,首先探讨了时域和频域上几种常用太赫兹波振幅成像方法所反映的样品信息及其特点,进一步使用提出的基于一定频率内平均吸收的太赫兹波振幅成像新方法对样品进行图像重构。实验结果表明这种新方法可以很好的反映样品的真实信息,反映了样品在一定频率范围内由于吸收而引起的效果的综合,与吸收系数和厚度相关,离散效应得到了很好的消除,相对几种常用的太赫兹波振幅成像方法能够得到更清晰的图像。此新方法尤其适用于结构简单的样品,能够成为几种常用振幅成像方法的有力补充。     

Ultrasonic flaw detection in NDE of highly scattering materials using wavelet and Wigner-Ville transform processing

In ultrasonic non-destructive evaluation of highly scattering materials the backscattering noise may attain peak values greater than the searched flaw pulse and the mean value of noise spectrum is very similar to the searched echo spectrum. Several specific methods have been proposed for the reduction of this type of noise, but the comparison of the performance of different methods is still an open problem. In this paper, we make a comparison among some methods based on simultaneous representations in time and frequency/scale domains of the ultrasonic traces. Synthetic and experimental traces are de-noised using a discrete wavelet processor with decomposition level-dependent threshold selection and a method that combines Wigner-Ville transform and filtering in the time-frequency domain. The results are comparatively evaluated in terms of signal to noise ratio and probability of detection.     

一种公共网络攻击数据挖掘智能算法研究

公共网络的开放性和自组织特性导致网络容易受到病毒干扰和入侵攻击,对攻击数据的准确高效挖掘能确保网络安全。传统方法采用时频指向性波束特征聚类方法实现攻击数据挖掘,在信噪比较低时攻击数据准确挖掘概率较低。提出一种基于自适应滤波检测和时频特征提取的公共网络攻击数据挖掘智能算法。首先进行公共网络攻击数据的信号拟合和时间序列分析,对含噪的攻击数据拟合信号进行自适应滤波检测,提高信号纯度,对滤波输出数据进行时频特征提取,实现攻击数据的准确挖掘。仿真结果表明,采用该算法进行网络攻击数据挖掘,对攻击数据特征的准确检测性能较高,对干扰的抑制性能较强,能有效实现网络安全防御。     

基于最优滤波器的强混沌背景中谐波信号检测方法研究

强混沌背景中的微弱谐波信号检测有重要的工程研究意义. 目前的检测方法主要是基于Takens理论的混沌相空间重构方法, 然而这些方法往往对信干噪比要求高, 且对高斯白噪声敏感等. 本文注意到混沌信号的二阶统计特性是不变的, 根据这个特点提出了一种基于最优滤波器的强混沌背景中的微弱谐波信号检测方法. 该方法首先构建一个数据矩阵, 在频域上对每个频率通道分别检测谐波信号, 从而将信号检测问题转化为最优化问题, 然后利用最优化理论设计滤波器, 使待检测频率通道的信号增益保持不变, 而尽量抑制其他频率通道的信号, 最后通过判断每一频率通道的输出信干噪比来检测谐波信号. 与传统方法相比, 本文方法有如下优点: 1)可以检测更低信干噪比下的微弱谐波信号; 2)可检测的信号幅度范围更大; 3)抗白噪声性能更强. 仿真结果证明了本文方法的有效性.     

基于Gammatone滤波器组时频谱和 卷积神经网络的海底底质分类*

为了有效利用海底底质信号完成海底底质的分类识别,本文提出一种将深度学习方法和底质信号相结合实现底质分类识别的方法。首先利用Gammatone滤波器组计算底质侧扫图像信号的时频谱,最后利用卷积神经网络(Convolutional Neural Networks, CNN)对得到的时频谱进行分类识别完成底质分类。实验结果表明该方法的底质分类准确率平均达到97.64%,相对于其他方法,分类性能更加优越;同时利用该方法分类海试数据,结果证明该方法具有一定的泛化能力。本文研究结果对实际的海底底质分类具有一定参考意义。