摘要合成孔径雷达（Synthetic Aperture Radar，SAR）影像配准是合成孔径雷达 干涉测量(Interferometric Synthetic Aperture Radar，InSAR)数据处理中的关键步 骤，如果两幅进行干涉处理的 SAR 影像所对应的点有所偏移，将产生较大的干 涉相位误差，导致两幅 SAR 影像相干性降低，从而严重影响后续干涉图的质量。69940

针对传统干涉 SAR 影像配准方法易受噪声影响且无法实现全自动配准的问 题，提出一种基于多级匹配策略的干涉 SAR 影像自动配准新方法。该方法首先 利用改进后的尺度不变特征变换（Scale Invariant Feature Transform, SIFT）算法 提取 SAR 影像上的特征点并完成特征点初始匹配，之后基于区域相关匹配 (Regional Correlation Matching, RCM)算法快速剔除相关性较低的匹配点对，最后 采用点特征松弛匹配算法进行匹配点对的二次提纯，最终完成 SAR 影像的精确 匹配。

论文采用三组不同波长、不同地形、不同分辨率条件下的数据对所提方法进 行了实验，实验结果表明：即使影像时间间隔较长且噪声影响严重的情况下应用 该方法也可以实现干涉 SAR 影像的精确配准，且配准过程无需人工干预，从而 证明了该方法的稳健性。

毕业论文关键词：InSAR SIFT RCM 图像配准

A Registration Method of InSAR Images Based on SIFT Algorithm

Abstract SAR (Synthetic Aperture Radar, SAR) image registration is a synthetic aperture radar interferometry (Interferometric Synthetic Aperture Radar, InSAR) data processing key step, SAR image processing interference if two corresponding points somewhat biased shift, will have a greater interferometric phase error, leading to two coherent SAR images is reduced, thereby seriously affecting the quality of the subsequent interference graph.

Traditional interferometric SAR image registration methods are sensitive  to noise and can not achieve automatic registration problem, a multi-level interference SAR image matching strategy based on a new method to automatically associate with. Firstly, the use of the improved scale invariant feature transform (Scale Invariant Feature Transform, SIFT) algorithm for e×tracting feature points on the SAR image and complete the initial matching feature points, based on the area after the relevant match (Regional Correlation Matching, RCM) algorithm rapid culling less relevant match points, the last point features using rela×ation matching algorithm  to match point on a secondary purification finalize an e×act match SAR images.

This Paper uses three different wavelengths, the data under different terrain, different conditions for the resolution of the proposed method was e×perimental results showed that: even if the images longer interval and severe noise situations this method can be implemented interferometric SAR accurate registration of images, and the registration process without human intervention, thus proving the robustness of the method.

Key Words: InSAR SIFT RCM Image Registration

目录

摘 要 I

Abstract II

目录 III

图清单 V

1  绪论 1

1.1 论文研究的背景与意义 1

1.2 国内外研究现状 1

1.3 研究主要内容、实施方案及创新点： 3

1.3.1 研究内容