小波域局部维纳滤波器估计误差分析及图像去噪

对小波域局部维纳滤波算法的估计误差进行了理论分析,推导了估计误差平方期望表达式,得到了一种观测系数局部方差估计的阈值.以此为基础,提出了一种小波域图像去噪算法.先对观测系数做阈值化处理,再进行局部自适应维纳滤波.实验结果表明,该方法提高了真实信号系数方差估计的准确度,在去噪性能上优于Mihcak等提出的LAWML算法.     

星间光通信系统中多样化波前畸变的小波重构

用小波变换方法对星间光通信中的多样化波前畸变进行重构,并分析了小波重构方法的两项参考指标:重构精度和重构算法复杂度。基于对这两项指标的综合考虑,提出一种最优小波基函数的选取方法。仿真数据表明,利用最优小波基函数选取方法对波前畸变进行重构,较之传统的Zernike多项式拟合方法,具有重构精度高且计算量小的特点。     

A Novel Method for non-contact measuring diameter parameters of wheelset based on wavelet analysis

A railway wheelset is subject to normal wear due to large part to friction contact between the wheelset and the rail. It is very important to measure the diameter parameter of wheelset for the safety of trains. The optoelectronic method is introduced and can dynamically measure the diameter parameter. Fast Fourier transform and wavelet analysis methods are used to denoise the detected signals of wheelset diameters. The results show that the wavelet analysis provides a much better way of denoising of the signals compared with the fast Fourier transform method. Comparisons of the diameter parameter measurements with the manual apparatus and the optoelectronic system are presented. The measurement accuracy of diameter parameter was within 1.0 mm. The root-mean-square error of the diameter is 0.51 and 0.15, respectively. The accuracy meets the measuring requirement of diameter parameters on line and shows that the system has better repeatability and reliability compared to the manual measurement.     

基于小波的多极化机载合成孔径雷达海面风向反演

为了实现从合成孔径雷达(SAR)图像本身提取高精度的海面风向信息, 提高SAR海面风场反演精度,研究了多极化机载SAR海面风向反演技术, 借助小波分析相对傅里叶分析和局部梯度更精细的时-频分析能力, 将二维连续小波变换与快速傅里叶变换(FFT)相结合,提出一种新的机载SAR海面风向反演方法. 为验证反演方法的有效性,通过海上同步飞行试验获取多极化机载SAR数据及同步调查船实测风向数据,用于反演试验的数据比对.采用本文提出的方法, 利用多种小波基对机载C波段SAR的同极化和交叉极化数据进行风向反演, 将反演结果与美国国家环境预报中心再分析资料以及调查船实测风向进行比对. 结果表明,本文提出的基于小波分析的海面风向反演方法适用于机载SAR探测数据, 反演精度优于二维FFT法和局部梯度方法;小波基的选择对反演结果影响较大, Mexican-Hat小波基是机载SAR海面风向反演的最优小波基, 且同极化与交叉极化机载SAR数据均可用于海面风向的反演.     

Adaptive edge detection method for images

The novel two-dimensional (2D) wavelet with anisotropic property and application of it has been presented. Wavelet is constructed in the polar coordinate system to obtain anisotropic properties. A novel edge detection method has been developed with the aid of this wavelet. This method detects gradient jump and than follows along this jump. In this way the number of calculation for edge localization is reduced. Moreover, the presented method is able to detect all edges in an image in multi-scale together with its spatial orientation. Proposed wavelet as well as edge extraction method seems to be new way to edge detection for an image.     

Improved holographic particle sizing by using absolute values of the wavelet transform

A new method for sizing particle from in-line particle holograms by using absolute values of the wavelet transform is proposed in order to improve accuracy in measurements. The proposed method provides direct calculation of the particle size by using spatial frequency information of a chirp signal at minima position of an envelope function. Simulation and experimental results are presented.     

轴对称体声振耦合的边界子波谱与有限元耦合方法

探讨了子波在Helmholtz积分方程及声振耦合中的应用,在建立了求解轴对称Helmholtz积分方程的子波谱方法的基础上,构造了轴对称子波谱与轴对称有限元的耦合方法,该方法可以处理轴对称问题的任意边界条件.进行了声振耦合问题的模态分析.     

利用深度神经网络和小波包变换进行缺陷类型分析

超声检测中对缺陷进行类型分析通常取决于操作人员对于特定专业知识的了解及检测经验,从而导致其分析结果的不稳定性和个体差异性。本文提出了一种使用小波包变换提取缺陷特征信息,并应用深度神经网络对得到的信息进行分类识别的方法。利用超声相控阵系统对于不锈钢试块上的通孔、斜通孔和平底孔进行超声检测,并对得到的超声回波波形按照新方法进行分析。实验结果表明,使用小波包变换后的数据进行分类识别能够在提高识别准确率的同时降低神经网络的学习时间,而使用深度神经网络相比通用的BP神经网络以可接受延长学习时间的代价提高了识别的准确率。采用新方法后,缺陷分类正确率提高了21.66%,而网络学习时间只延长了91.9s。在超声检测中使用小波包变换和深度神经网络来对于缺陷进行类型分析,能够排除人为干扰,增加识别准确率,对于实际应用有着极大的意义。      

基于多小波变换的图像去噪新方法

离散多小波在图像去噪和图像融合方面有突出的效果.利用GHM多小波,对图像去噪中阈值的选取给出了具体的方法,对单小波与多小波去噪方法进行了比较.结果表明,多小波在图像去噪的效果上明显优于单小波.     

Comparison of the efficiencies of Fourier and wavelet expansions in passive acoustic thermal tomography

A comparison of the accuracy achieved in temperature-distribution reconstruction by the Fourier and wavelet expansions in passive acoustic tomography is carried out. Since the use of focused ultrasound in hyperthermia leads to local heating of the tissue, the wavelet representation of such temperature distributions with local fluctuations is more compact than their Fourier representation. It is demonstrated that the compactness of the wavelet representation provides an opportunity to reduce considerably the number of unknown quantities in solving the inverse problem of acoustic thermal tomography and to increase the accuracy of the temperature-distribution reconstruction in comparison with the case of using the Fourier representation. A method for choosing a compact wavelet basis for temperature distributions obtained in the hyperthermia process is proposed.     

小波聚类方法和近红外光谱技术用于药片种类判别

对310个药片样品的近红外光谱数据进行了聚类分析。首先使用小波变换对光谱数据矩阵进行多尺度分解,在进行有效压缩之后,采用经典分类方法对合适选取的小波系数组合进行聚类分析,提出了小波聚类方法。该方法分别用于实验室药片、中试药片和规模生产药片样品的分析,按药片样品的组成得到4个类别。结果表明,对实验室药片和中试药片样品分类的精确度均达到100%;对于规模生产药片的分类,共120个样品中只有1个样品被错误划分,精确度也高达99.2%。近红外光谱技术结合小波聚类方法的聚类性能是令人满意的,相比经典聚类分析,更加快速、易于使用,对制药行业药片质量以及成本控制均有积极作用。     

基于小波变换和数学形态学的孔径测量研究

针对视觉测量中硬盘圆孔孔径容易受到圆孔周围局部强反射、外部噪声的影响和测量精度不高的问题,提出一种基于小波变换、数学形态学和机器视觉相结合的圆形零件孔径测量方法。对摄像头采集圆孔图片,通过小波变换分解出高低频图像,利用小波变换对高频部分进行边缘提取,数学形态学对低频部分进行边缘提取,然后进行小波边缘融合,获取有效的圆孔边缘。利用最小二乘法对边缘进行孔径尺寸计算,经实验验证,该算法与其他算法相比,在有效地保留圆孔周围边缘信息的情况下,对周围噪声进行了有效的抑制,检测精度为0.01 mm以内,实验结果表明此方法简单易行,且具有较高的精度。     

基于人的视觉特征的双正交小波基的构造

构造了一个与人的视觉系统模型的调制转移函数相近似的紧支撑双正交小波基。文中采用遗传算法，寻找一个紧支撑双正交小波函数，使它与调制转移函数在频域中相近似。同时将所构造的双正交小波基用于图像压缩，并将它与其它双正交小波进行比较。实验结果表明，所构造的紧支撑双正交小波基与Daubechies的9／3小波相比较有更高的压缩比，与9／7小波相比压缩性能相当，但计算量更小，它更适合于纹理图像的压缩。     

光谱信号的小波去噪新技术

光谱分析中,噪声的存在常影响分析的准确度和检测限。现有滤波方法在光谱信号除噪方面有种种缺陷。文章充分利用小波在信号处理方面的优良特性,提出了Mexican Hat小波滤波算法,它选用Mexican Hat小波函数构造滤波项,利用滤波项与原始信号作用,从而实现信号与噪音的分离。该方法无论对低频或高频信号均适用,除噪完全,即使对信噪比为1的高噪声信号也能获取满意的处理结果。运用这一技术处理光谱信号,简单快速、结果可靠,处理后峰位置、峰高、峰面积误差分别小于0.2%,3.2%,1.1%。大量实验表明,本方法能有效提高光谱分析的准确度。     

A load identification method based on wavelet multi-resolution analysis

Load identification, as a kind of indirect identification method, uses system characteristic and responses to calculate loads. A method based on wavelet multi-resolution analysis is proposed in this paper. By wavelet decomposition and transform at certain resolutions, the proposed method transforms the convolution relation between responses and loads in time domain into the linear multiplicative relation between system responses and wavelet responses in the wavelet domain. Loads can be identified as long as the linear multiplicative relation is solved. Qualitative and quantitative rules are proposed for selecting parameters that affect the accuracy of the proposed method, and are illustrated via numerical investigations. The method is illustrated by a multi-input-multi-output numerical simulation. A multi-input-multi-output laboratory experiment is performed to compare the proposed method with the frequency method on the identification ability.     

基于改进小波阈值的量子点荧光光谱降噪研究

使用光谱仪采集到的信号难免受到不同噪声源的影响。为了提高光谱信号解析的精准度,通过分析小波应用于信号降噪的原理以及经典的软、硬阈值降噪法存在的缺陷,提出了一种改进的阈值降噪法。该方法既克服了硬阈值法产生间断点,软阈值法产生恒定偏差的缺陷,又尽量地保留了有用信号。实验选用的小波基函数为SymletsA,分解层数为4,结合Birge Massart策略模型确定的分层阈值对硒化镉量子点荧光光谱信号进行降噪处理。结果表明,与经典的软、硬阈值降噪法相比,通过改进阈值降噪法得到重构信号的信噪比（SNR=47.550 2）、能量占比（PER=0.973 3）和均方误差（MSE=149.421 3）均有提高和改善。     

基于在线小波支持向量回归的混沌时间序列预测

混沌时间序列预测是非线性动力学研究中一个十分重要的问题，支持向量回归方法为其提供了一种有效的解决思路.通过分析新样本加入训练集后支持向量集的变化情况，建立了一种混沌时间序列预测的支持向量回归算法，具备了在线学习的特点.同时，针对混沌信号提出了一种满足小波框架的小波核函数，它不但能以较高的精度逼近任意函数，而且适合于混沌信号的局部分析，提高了支持向量回归的泛化能力.最后就Mackey-Glass混沌时间序列在线预测问题进行了大量仿真.结果表明，本文算法与现有的算法相比具有训练时间短、预测精度高等特点，有一定 关键词： 混沌时间序列 支持向量回归 在线学习 小波核     

The Paul wavelet algorithm: an alternative approach to calculate the refractive index dispersion of a dielectric film from transmittance spectrum

A new application of the Paul wavelet algorithm was presented to determine the refractive index dispersion of a dielectric film from transmittance spectrum in the visible and near infrared region. The developed algorithm was tested by simulated data and experimentally applied to a sample of mica. The obtained refractive index dispersion determined by the Paul wavelet algorithm was compared with the refractive index values determined by the envelope and fringe counting methods, and also with the established result. It was shown that the degree of the Paul wavelet has a major effect on the outcome of a refractive index determination. The noise immunity of the presented method was shown by the simulation study.     

结合小波分析法与导数光声光谱技术测量弱光谱信号的实验研究

结合导数光声光谱技术与小波分析方法,精确测量了光声光谱中的弱光谱信号.首先利用自己设计的仪器装置实现光声光谱的一阶导数,在此基础上,根据导数光声光谱的数据选择合理的分析小波,将光声光谱信号分解为不同频率信号的叠加,被分解的信号满足线性性质且原信号的峰值信息保持不变,由频率的差异可区分出光声信号和噪声信号,从而提取出光声光谱中的弱光谱信号.结合物理方法与软件方法的分析结果,准确地测量了氙灯的输出光声光谱中3个不明显的弱峰值,位置分别为699.7nm、753.4nm和776.5nm.该方法可以更准确地提取出光声光谱的峰值信息,有效地提高了光声光谱的测量精度,为光声光谱分析法在生物医学及化学分析中的应用提供了一种更精确的分析方法.     

基于小波神经网络的振动速度传感器幅频特性补偿研究

为了实现超低频振动速度测量, 提出补偿其幅频特性的小波神经网络方法.该方法以振动速度传感器动态实验数据为基础, 通过小波神经网络训练来确定传感器幅频特性补偿网络.介绍振动速度传感器幅频特性补偿原理, 分析网络的拓扑结构, 给出网络参数训练和初始化方法.采用引入动量项的最速下降法训练网络权值、尺度因子和平移因子, 将小波网络参数的初始化与小波类型、小波时频参数和学习样本等联系起来.结果表明, 采用小波神经网络进行振动速度传感器幅频特性补偿具有良好的鲁棒性,并能实现在线补偿,网络训练的速度和精度优于同等规模的BP网络,在测试领域有重要的实用价值. 关键词： 振动速度传感器 小波神经网络 幅频特性 补偿