基于实测SAR图像的杂波特性研究和图像重构

发布时间：2018-02-26 15:18

本文关键词： 地杂波实测SAR图像幅度分布特性空间相关性杂波仿真　出处：《西安电子科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：在进行复杂环境中目标的检测和识别时,雷达接收到的回波信号中不仅包含目标的信息,同时也包含了目标所处地面环境的散射回波,即地杂波。雷达目标环境的复杂多变性,决定了地杂波对于雷达目标探测的重要意义,为了提高目标识别的准确性,必须对特定的地杂波特性进行分析和研究,才能够有效的抑制杂波。论文结合实际背景的需求,基于实测农田和水泥的合成孔径雷达(SAR)图像,深入分析了其地杂波的统计分布特性和空间相关性。论文主要工作如下:1.分别从地海杂波的实验测量、基于杂波散射机理的后向散射以及基于统计的杂波仿真三个方面,总结阐述了国内外目前对杂波的研究现状以及发展趋势。2.讨论了地杂波的雷达散射机理以及影响雷达散射截面的条件,阐述了几种常用的地面环境电磁散射分析的数值方法和解析方法。介绍了瑞利分布、对数正态分布、韦伯尔分布、K分布四种统计分布模型,给出了四种模型的概率密度随分布参数的变化情况,并列出各个分布函数相应的参数估计的方法。3.读取得到实测图像的数据,对所获得的数据进行统计分析,并且经过参数估计算法得到各个统计模型的分布参数,将各个经验模型的概率密度与原始图像的幅度分布概率密度进行拟合。学习了两种经常用到的拟合优度检验方法:K-S检验和卡方检验方法。并用K-S检验方法对拟合进行检验,得到了对数正态分布模型为最优模型。根据图像数据的相关性,分析了杂波数据的空间相关性。4.比较了两种典型的杂波模拟方法,即零记忆非线性变换法(ZMNL)和球不变随机过程法(SIRP)的优缺点。阐述了仿真过程中高斯白噪声的生成以及滤波器的获取方法。根据估计得到的分布参数,用对数正态分布(Log-normal)分布的ZMNL仿真方法,仿真得到农田杂波和水泥杂波的分布图,并将仿真得到的杂波图的概率密度分布与原始图像的对比。给出了通过ZMNL仿真方法仿真得到的杂波回波数据。5.通过对合成孔径雷达的基本理论的学习,对比了两种经典的成像算法线性调频算法和距离多普勒成像算法,结合仿真得到的杂波回波数据,采用距离-多普勒成像算法对杂波仿真结果进行成像。并将成像结果与原始图像以及杂波仿真图进行重构对比。
[Abstract]:In the detection and recognition of target in complex environment, the echo signal received by radar contains not only the information of the target, but also the scattering echo of the ground environment of the target, that is, the ground clutter and the complex variability of the radar target environment. The importance of ground clutter for radar target detection is determined. In order to improve the accuracy of target recognition, it is necessary to analyze and study the specific ground clutter characteristics in order to suppress clutter effectively. Based on the measured synthetic Aperture Radar (SAR) images of farmland and cement, the statistical distribution characteristics and spatial correlation of ground clutter are analyzed in depth. The main work of this paper is as follows: 1. Backscattering based on clutter scattering mechanism and clutter simulation based on statistics. The research status and development trend of clutter at home and abroad are summarized. 2. The radar scattering mechanism of ground clutter and the conditions affecting radar cross section are discussed. In this paper, several numerical and analytical methods for electromagnetic scattering analysis in ground environment are described. Four statistical distribution models, Rayleigh distribution, logarithmic normal distribution and Webster distribution K distribution, are introduced. The variation of probability density with distribution parameters of four models is given, and the methods of parameter estimation for each distribution function are listed. 3. The data obtained from the measured images are read and the obtained data are analyzed statistically. And through the parameter estimation algorithm, the distribution parameters of each statistical model are obtained. The probability density of each empirical model is fitted to the probability density of the amplitude distribution of the original image. Two commonly used methods of goodness of fit test: K-S test and chi-square test are studied. The fitting is tested by K-S test method. The logarithmic normal distribution model is obtained as the optimal model. According to the correlation of image data, the spatial correlation of clutter data is analyzed. 4. Two typical clutter simulation methods are compared. The advantages and disadvantages of ZMNL (Zero memory nonlinear Transformation) and SIRP (spherical Invariant Stochastic process) are discussed. The generation of Gao Si white noise and the acquisition of filters in the simulation process are described. According to the estimated distribution parameters, Using the ZMNL simulation method of log-normal distribution, the distribution maps of farmland clutter and cement clutter are obtained by simulation. The probability density distribution of the simulated clutter is compared with the original image. The clutter echo data. 5. The basic theory of synthetic Aperture Radar (SAR) is studied. Two classical imaging algorithms, linear frequency modulation (LFM) algorithm and range Doppler imaging algorithm, are compared. The range Doppler imaging algorithm is used to image the clutter simulation results, and the imaging results are compared with the original images and the clutter simulation images.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.52

【参考文献】