全极化合成孔径雷达影像地形纠正及其在雪冰制图中的应用

发布时间：2018-05-20 20:15

本文选题：全极化合成孔径雷达 + 地形纠正　；参考：《中国图象图形学报》2014年01期

【摘要】：目的针对合成孔径雷达(SAR)影像由于地形起伏引起的图像畸变问题,提出了基于相干矩阵的全极化SAR影像地形纠正算法,并运用于雪冰制图。方法该方法首先采用距离多普勒模型建立SAR成像几何模型;然后利用全极化Cloude特征分解方法对全极化SAR图像进行融合,将融合后的SAR图像与模拟图像进行配准提高SAR影像几何定位精度;最后利用投影面积归一化和极化方位角移动补偿技术对地形引起的辐射畸变进行纠正。结果采用中国长江源区南部唐古拉山中段冬克玛底冰川区域的C波段Radarsat-2全极化SAR数据进行验证,配准模拟SAR和原始SAR影像的控制点方位向和距离向的均方根误差(RMSE)分别为7.765像素和14.586像素;经过地形纠正后的地物分类精度达80%以上。结论实验结果表明:1)该方法能够有效消除SAR影像中几何和辐射畸变的影响;2)地形纠正后的SAR数据在雪冰制图中具有可行性。
[Abstract]:Aim to solve the problem of image distortion caused by terrain undulation in synthetic aperture radar (SAR) images, a full polarimetric SAR image terrain correction algorithm based on coherent matrix is proposed and used in snow ice mapping. Method this method first sets up a few models of SAR imaging using distance Doppler model, and then uses full polarimetric Cloude characteristic decomposition square. All polarization SAR images are fused, the fused SAR image is registered with the simulated image to improve the geometric positioning accuracy of the SAR image. Finally, the projection area normalization and polarization azimuth motion compensation technique are used to correct the radiation distortion caused by the terrain. The C band Radarsat-2 fully polarizing SAR data of the bottom glacier area is verified. The registration of the azimuth and distance root mean square error (RMSE) of the control points of the simulated SAR and the original SAR images is 7.765 pixels and 14.586 pixels respectively. The classification accuracy of the terrain after the topographic correction is more than 80%. Conclusion the experimental results show that: 1) the method can effectively eliminate it. In addition to the influence of geometric and radiative aberrations in SAR images, 2) SAR data after topography correction are feasible in snow and ice mapping.
【作者单位】：中国科学院遥感与数字地球研究所;中国科学院大学;
【基金】：国家自然科学基金项目(41201447) 国家高技术研究发展计划(863)基金项目(2011AA120403) 国家重点基础研究发展计划(973)基金项目(2010CB951403)
【分类号】：TN957.52

【参考文献】