干涉条纹图处理的等值线窗口滤波和摄像测量在国防试验的应用

对图像进行滤波或称平滑是干涉条纹图像处理中的一项重要工作。一幅未经处理的原始图像或多或少存在着不同程度的噪声干扰,特别是散斑和Insar干涉条纹图,信噪比很低,难以处理。本文针对光测力学中光学干涉方法得到的条纹图,提出了一种新的滤波方法条纹等值线窗口滤波,并对这种滤波方法进行了研究讨论,提出了几种确定等值线窗口的不同方法。这种滤波方法根据干涉条纹图不同断面上灰度分布的不同特点,选择沿条纹走向的条纹等值线窗口进行滤波,在最大消除条纹图噪声的同时,也能保证对条纹损伤最小。摄像测量技术正在迅速发展和得到广泛应用,在国防试验和航天飞行任务中发挥着不可替代的作用。本文也介绍了作者近年来在该领域所做的应用研究,包括在火箭、导弹发射试验,冲击、碰撞等过程中的动态目标运动测量;针对航天航空和力学工程领域的视频图像实时分析;飞行器三维运动测量;基于投影条纹的物体三维形貌精密测量方法研究;以及对接航天器位置和姿态的实时测量。在这些应用中都实现了高精度测量。     

基于多通道滤波技术的散斑条纹图正则化方法

在光测力学中,条纹图正则化是进一步提取条纹位相信息的重要基础。对于散斑条纹图来说,正则化的过程还需能够有效地抑制散斑噪声。本文提出一种基于多通道滤波技术的散斑条纹图正则化方法。通过使用多个Gabor滤波通道对散斑条纹图进行滤波,并且引入与条纹对比度有关的权重因子把各个滤波通道的滤波结果叠加起来,最终得到滤除了散斑噪声的正则化条纹图。数值模拟和实验结果表明,该方法在将散斑条纹图正则化的同时能够有效地滤除条纹图中的散斑噪声,为进一步提取条纹位相奠定了良好的基础。     

干涉条纹图中散斑噪声之滤除

散斑在光学干涉计量中被用作信息载体，但同时它也是严重的噪声。介绍了一种简单有效的滤除散斑噪声的方法，即在加窗富里叶变换域内进行掩模和阈值处理。同富里叶变换和正交小波变换相比，加窗富里叶变换有如下两方面的优点：加窗富里叶基和条纹的高度相似性使得信号与噪声容易分离，而加窗富里叶变换的冗余性则可削弱掩模和阈值处理对信号谱的破坏。此方法的有效性由数值模拟和实验的例子证实。此方法亦可滤除强度型条纹或位相型条纹中其它类型的噪声，如加型噪声。     

基于加窗傅立叶变换的SR-CT滤波反投影重建

SR-CT(Synchrotron Radiation Computed Tomography同步辐射X射线计算机断层扫描)实验中,原始投影图像存在噪声,从而影响了滤波反投影重建图像的质量.为提高重建图像质量,提出了一种基于加窗傅立叶变换滤波改进重建图像质量的方法.通过对原始投影图像噪声和滤波反投影算法的讨论,对利用该方法提高重建图像质量的可行性进行了分析.利用加窗傅立叶滤波方法对两组SR-CT实验得到的投影图像分别进行了处理,抑制了原始投影图像的噪声,提高了重建图像的质量.实验的结果验证了该方法的有效性.     

区域密度滤波及其在数字散斑条纹处理中的应用

利用统计学原理对数字错位散斑条纹的形成进行了系统分析，并针对散斑条纹图所特有颗粒性强的特点，提出一种新的空域滤波方法－区域密度滤波。该方法能更有效地抑制散斑图中的高频噪声，也使条纹图更平滑，并且细节得到了最大程度的保存，避免了有些空域滤波法因误差积累而导致条纹移动的问题，进一步保证了相移技术和神经网络系统的计算精度。     

数字剪切散斑相关条纹图的形成及其分析

本文提出了用线性相关计算形成高可见度的数字剪切散斑相关条纹图的方法，为相关条纹图的自动分析奠定了良好的基础。文中还推导了附加位相与物光反射镜转角的关系，使精确地计算附加位相成为可能。同时还介绍了任意附加位相的三步相移技术。最后给出了用线性相关计算法形成的相关条纹图及计算出的位相图。     

基于小波分析的条纹图滤波方法

本文提出了一种新颖的干涉条纹图非线性滤波方法,可以选择性地在频域图像的不同区域选用不同的滤波方式,在滤掉图像中大部分散斑噪声的同时,能够减少信息的损失,使图像的内部边界仍然保持清晰。文中首先简介了正交小波变换的原理,然后介绍了去除散斑噪声的具体算法,最后给出了计算机模拟去噪声的结果     

光载波条纹图的锁相处理

本文提出了光载波条纹图的锁相处理方法,以对光载波条纹图作精确定位。给出了适合于计算机处理的离散递归实现步骤。在非线性模型下对定位精确度作了详细分析。实验结果表明,当暂态影响衰减后,能对条纹图进行精确定位而无稳态误差,对断裂条纹的大范围补偿也取得了较好的效果。     

Fourier and wavelet analyses for fatigue assessment of concrete beams

We investigate damage detection in a simply-supported pre-stressed beam. A crack was propagated by fatigue loads, which were applied up to two million cycles. Both fast Fourier transform (FFT) and continuous wavelet transform (CWT) are used in the analysis of the structural response to impulse loads. The acceleration response of the full-scale beam was measured each time a certain number of cycles of fatigue loads were applied. The results of this study show that both methods can clearly identify the crack growth induced by fatigue loads. The natural frequencies found by FFT are sensitive to the crack progression. The results from the CWT analysis show a clear difference in structural responses between the initial and damaged states of the structure. The response accelerations are de-noised by a soft-thresholding method before they are analyzed by CWT. In addition to the frequency components, the CWT shows the moment in time when particular frequencies occur. Therefore, wavelet analysis has the potential of becoming an effective tool for damage detection and health monitoring of structures for which the natural frequencies are irregularly changing. As the crack grows, the magnitude of ridges obtained by CWT analysis decreases significantly, which indicates the reduction in structural stiffness.     

Fourier analysis of several finite difference schemes for the one‐dimensional unsteady convection–diffusion equation

This paper reports a comparative study on the stability limits of nine finite difference schemes to discretize the one‐dimensional unsteady convection–diffusion equation. The tested schemes are: (i) fourth‐order compact; (ii) fifth‐order upwind; (iii) fourth‐order central differences; (iv) third‐order upwind; (v) second‐order central differences; and (vi) first‐order upwind. These schemes were used together with Runge–Kutta temporal discretizations up to order six. The remaining schemes are the (vii) Adams–Bashforth central differences, (viii) the Quickest and (ix) the Leapfrog central differences. In addition, the dispersive and dissipative characteristics of the schemes were compared with the exact solution for the pure advection equation, or simple first or second derivatives, and numerical experiments confirm the Fourier analysis. The results show that fourth‐order Runge–Kutta, together with central schemes, show good conditional stability limits and good dispersive and dissipative spectral resolution. Overall the fourth‐order compact is the recommended scheme. Copyright © 2001 John Wiley & Sons, Ltd.     

Modelling of the diffusion of fluid particles in turbulent flows approaching a circular cylinder by random Fourier modes and random flight

To investigate the diffusion of fluid particles around a cylinder in a turbulent flow, we have developed two new types of model for simulating the trajectory of particles:(1) a model combining random Fourier modes and random flight (RF); (2) a pure kinematic simulation (KS) by random Fourier modes. In model 1 the large-scale turbulence is simulated by a sum of random Fourier modes varying in space and time, and the small-scale random motion of particles is simply modelled by an Itô type of stochastic differential equation with a memory time comparable to the Lagrangian time scaleT _s ^L of the small-scale motion. In model 2, both large- and small-scale turbulence is simulated using random Fourier modes. The change of turbulence around the cylinder is modelled by rapid distortion theory (RDT), although the small-scale motion of particles in the RF model is simply assumed to keep the homogeneous random behaviour. These models give very similar and realistic trajectories showing rapid changes of direction due to the small-scale motion.