【作者】 杨夏；

【导师】 于起峰；

【作者基本信息】 国防科学技术大学 ， 航空宇航科学与技术， 2007， 博士

【摘要】 ESPI(Electronic Speckle Pattern Interferometry,ESPI)测量技术是一种对光学粗糙表面进行无损全场测量的技术,被广泛地应用于振动、位移、应变和医学诊断等各方面的测量中。干涉合成孔径雷达技术(Interferometric Synthetic AperatureRadar,InSAR)是在合成孔径雷达(SAR)的基础上发展起来的新技术,主要用于获取大面积高精度数字高程图(DEM)和进行高精度的地表变化监测。ESPI和InSAR是比较相似又互相区别的两种技术,它们都以干涉条纹为主要研究对象,力争得到高质量的相位条纹图,进而获取高精度的全场相位并根据相位求解待测物理量。所以条纹图处理是这两种测量技术的关键步骤之一,条纹图处理方法的优劣将直接决定这两种技术的测量精度和实用能力。本文在深入研究ESPI条纹图处理方法的基础上,提出如下创新点:(1)深入研究了常用的条纹方向求取方法——平面拟合法与梯度法的优缺点,并将两者结合使其互补,得到了更为精确的条纹方向。提出了另一种求取条纹方向的累积平方差方法,并用实验证明了该方法优于已有方法。在认真分析ESPI与InSAR条纹图特点的基础上,分别提出了两种ESPI条纹图条纹密度的估计方法——累积灰度差法和条纹中心线法以及两种InSAR干涉条纹图条纹密度的估计方法——基于二值化条纹的方法和基于条纹梯度的方法。条纹方向和条纹密度是本文后续研究的重要基础。(2)提出了两种确定等值线窗口的新方法:一种是方向与梯度结合的确定法,一种是中心线等分法。通过实验对比,证明了两种新方法所得到的等值线窗口精度都优于原有方法。等值线窗口也是本文对条纹图进行处理的一个重要基础。(3)通过对ESPI条纹图条纹密度的估计,结合等值线窗口滤波法,实现了用于ESPI条纹图的自适应等值线窗口滤波方法。该方法对于ESPI条纹的走向变化和密度变化都具有了一定的自适应能力,能在较好保持条纹信息的前提下更好的滤除不同密度条纹的噪声。在此基础上,本文还利用估计的条纹密度对等值线相关干涉法(CCI,Contoured Correlation Interferometry)进一步发展,形成了窗口尺寸可以自适应的CCI方法,提高了CCI方法的适应性和实用性,使相关方法对于宽条纹和窄条纹能同时得到较好的条纹结果。(4)提出了用于InSAR干涉条纹图的等值线窗口滤波法并对其进行了深入研究,从理论和实验上都证明了该滤波方法优于正余弦滤波法和Lee滤波方法。结合本文前面的两种估计InSAR条纹密度的方法,实现了用于InSAR干涉条纹图的自适应等值线窗口滤波方法,在较好保持InSAR干涉条纹信息的前提下进一步增强了噪声滤除的能力。(5)提出了基于条纹切面角的条纹中心线确定方法和符号图确定方法,使抗噪声能力和结果精度优于已有方法。还根据条纹方向对图像边缘的中心线进行了补充,并实现和改进了从单幅ESPI条纹图提取相位的正则化方法与积分变换法。研究了快速质量引导解缠方法,大大加快了质量引导解缠方法的处理速度,这对于大量InSAR数据的处理具有重要意义。提出了了基于等值线判定的质量图引导解缠法,提高了质量引导法的解缠可靠性。更多还原

【Abstract】 Electronic Speckle Pattern Interferometry (ESPI) is a whole-field non-destructive technique to measure optical rough surfaces. It is widely used in the measuring of vibration, displacement, strain and medical diagnosis. Interferometric Synthetic Aperture Radar (InSAR) is a new technology developed from the technology of Synthetic Aperture Radar (SAR). And it is mainly used to obtain the Digital Elevation Model (DEM) in a vast area precisely and to monitor minute changes of the ground surface. These two techniques are closely related although different. Both of them take interferograms as their main object of research and aim to get high quality fringes so as to get good whole-field phases, based on which the wanted physical parameters will be calculated. So fringe processing is a key step for both techniques by influencing their precisions and applications. Based on a thorough exploration into these two kinds of fringe processing techniques, this dissertation presents its originality in the following aspects:(1) The merits and shortcomings of two fringe-orientation obtaining methods, including the plane-fit method and the gradient method, are well studied. And they are combined to overcome their disadvantages so as to get more precise orientation results. The accumulated variance method is also put forward to calculate fringe orientations. Based on the analyzing of ESPI and InSAR fringe characters, this dissertation proposes two ESPI fringe-density estimating methods including the accumulated difference method and the fringe skeleton method, and two InSAR fringe-density estimating methods including the binary fringe method and the gradient-based method. Fringe orientation and fringe density are two kinds of important information for the following researches in this dissertation.(2) Two methods of defining a contoured window are carried out. One is the method based on orientation and gradient and the other is the skeleton-distance averaging method. Experiments show that the contoured windows obtained with these two methods are more precise than those obtained with original ones. Contoured window is also important for the following researches.(3) Based on the contoured window filter (CWF) and the estimations of fringe orientation, an adaptive contoured window filter (ACWF) is realized for ESPI interferograms. The ACWF can shift its shape to the changes of fringe orientation and alter its size to the changes of fringe density. In this way, noise can be depressed better while the fringes being kept well. Combined with the adaptive contoured window, the contoured correlation interferometry (CCI) method is also developed by making its window size adaptive. In this way, the CCI method can be more adaptive and applicable and can get better correlation results for both wide and thin fringes at the same time. (4) The application of CWF for InSAR fringes are proposed and well studied. This filter is proved theoretically and experimentally to be better than both the sine/cosine filter and the Lee filter. Combined with CWF and the estimation of InSAR fringe density, adaptive window filter for InSAR is achieved so as to filter noise better while keeping fringes well.(5) This dissertation also proposes a method to locate skeletons and calculate the sign map according to the fringe gradient angles. Both the results and the resistance to noise are better than those of the original methods. A method of completing skeletons along fringe orientations is carried out, and the fringe-normalizing method and the quadrature transform method are refined to calculate phase from one filtered ESPI image. A fast quality-guided phase unwrapping method is researched. This technique shortens the processing time in a great deal and is significant when dealing with vast InSAR data. A method of quality-guided phase unwrapping based on contoured window judge is also proposed to enhance the unwrapping reliability.更多还原