【作者】 任璐；

【导师】 保铮；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 2005， 硕士

【摘要】 合成孔径雷达(SAR)能够全天时、全天候对远距离目标进行二维高分辨成像,在军用和民用的等众多领域都具有广泛的应用。随着对成像雷达研究的不断深入,人们对成像的质量,如图像的分辨率,可判读性及视觉效果,提出了更高的要求。为此,本文从理论上具体研究了如何通过成像后处理的方法进一步提高SAR 图像的质量,并通过实测数据验证了所给算法的有效性。论文的第一章描述了SAR 成像模型及原理,回顾了SAR 图像后处理的发展历史并介绍了其研究现状,最后概括了全文的主要内容。第二章研究了SAR 图像中一类动目标的自聚焦。并基于SAR 成像中散射点模型的参数估计,提出了一种适用于经过距离走动校正之后,在方位向上仍存在两次及两次以上相位误差的动目标的自聚焦算法,通过交替迭代运算,可以通过较少的距离单元实现较为精确的误差补偿。经过实测数据验证,该算法可以获得理想的自聚焦。第三章研究现代谱估计中的超分辨成像算法。主要分析了APES 和CAPON 一类算法的原理,并对这些方法利用模拟数据及实测数据进行了数值试验,提高了复图像的质量。目前由于仍然没有一个统一的关于合成孔径雷达成像超分辨率概念上的解释和评判标准,而谱估计方法往往需要对信号做出一定的假设。究竟这些方法能否广泛的应用到实际的雷达成像处理中,还值得进一步的探讨。第四章研究了SAR 图像的相干斑抑制。分析了相干斑形成的机理及滤波方法的研究现状,并提出了一种基于空间尺度相关法的SAR 图像自适应去相干斑算法。通过该滤波算法处理,SAR 图像不仅可以获得令人满意的去斑效果,并且较好的保持了细节和纹理信息。实测数据证明,该方法具有较强的适用性。第五章对全文工作进行总结,并指出有待进一步研究的问题。更多还原

【Abstract】 Synthetic Aperture Radar (SAR) has an all-weather and day/night capability togenerate a two dimensional high-resolution image of the targets over a long range, so itis widely used in both civilian and military fields. With the further investigation inimaging radar, people demand to obtain better qualities of SAR images, includinghigher resolution, easy readability and clear visual effect. According to this, thedissertation mainly investigates how to improve the quality of SAR images by themeans of post imaging processing methods from the theoretical perspectives.Meanwhile some real data are provided to test our algorithms’effect.The first chapter is mainly associated with the theories of SAR imaging based onimaging model, then reviews the development history of SAR post imaging process,introduces the existed research at present, and at last concludes the whole contents ofthe dissertation.The second part discusses the autofocus technique about one kind of moving targetsin SAR images. Based on the parameters estimation of SAR scene’s scatters, a newautofocus algorithm is proposed for moving targets, which have no range alignment butnearly two orders of polynomial phase errors exist in azimuth. Through alternateiteration, this approach can compensate the exact errors with less range cells.The third chapter studies some super-resolution algorithms in modern spectrumestimation. APES and CAPON algorithms are analyzed and used to make experimentswith SAR images. Up to now, no general comments and standards are given to explainthe concepts of super-resolution of radar imaging, so these methods’values in practicalapplications are still worth further discussions.The fourth part discusses the speckle suppression in SAR images. The mechanism ofspeckle produced is studied and some existed filtering approaches are discussed. Then anew adaptive filtering algorithm is proposed to remove speckle in SAR images based onthe inter-scale correlation in wavelet domain. The real data prove that this approach hassuperior performance than other conventional methods.The last chapter concludes all the contents of this dissertation and points out theproblems left to be further discussed.更多还原