基于影像模拟的SAR几何校正准自动方法

A Quasi-automatic Rectification Method of SAR Image Based on Image Simulation

合成孔径雷达影像的几何校正是许多微波遥感应用中必须解决的问题 ,当无法获得准确的轨道数据时 ,这个问题变得非常困难。本文提出了利用合成孔径雷达模拟影像进行准确的几何校正的原理及方法 ,并以RADARSATSAR影像进行了实验。该方法利用不准确的轨道数据及数字高程模型生成模拟的SAR影像 ,用影像匹配的方法自动获取模拟影像与真实影像之间同名点的坐标差值 ,而这个差值信息正好提供了对不准确轨道数据的控制。本方法无须地面控制点 ,并基本上可以自动进行 ,是解决目前SAR几何校正问题的有效方法。

SAR image usually becomes very difficult when precise orbit data of SAR platform is not available. A SAR rectification method based on image simulation is proposed in this paper. The simulated SAR image is generated by means of Digital Elevation Model (DEM) and the known orbit data, then image matching is performed between the simulated and the real SAR image to identify some pairs of corresponding points. The position differences of these corresponding points are caused by the imprecision of the orbit data, and can be used to correct the rectification residuals. In this paper, the common SAR image simulation algorithms are introduced briefly, and then the algorithm used to match real to the simulated image is presented, as well as the SAR rectification method based on simulated image. At last, an experiment on the proposed rectification method is made, in which a RADATSAR SGF image is rectified precisely by means of simulated SAR image. Moreover, except the image matching stage, all the procedures of this rectification can be done automatically, which is significant to the large-volume SAR image processing.