中文摘要：

针对新型高分辨率雷达卫星TerraSAR—X立体像对，本文提出采用基于RPC的平差模型，通过少量的地面控制点来拟合因传感器不稳定、平台星历数据不精确及测距误差引起的影像几何畸变，从而达到精确定向目的。为验证RPC平差模型的适用性，通过在立体成像区域均匀布设人工角反射器点的方法验证其模型精度，并评估了其三维定向平差后的精度。实验结果表明RPC平差模型能很好地适用于TerraSAR—X立体数据，并可获得4点控制条件下，平面0．166m和高程0．241m的精度，说明利用RPC平差模型可以对TerraSAR—X数据的立体定向达到很好的精度。

英文摘要：

Recently, terrain information extraction from satellite SAR data has been an active research topic enjoying the advantages of radar systems and of digital image processing： all-weather, night and day operation. The use of stereoscopic SAR images for the generation of DEMs has been an important application of SAR data. In this paper, an image based transformation to correct SAR image geometric distortions due to the sensor instability, platform ephemeris instability and target ranging inaccuracy, using a small number of control points in cooperation with the Rational Polynomial Coefficient （RPC） model to improve the space intersection accuracy was proposed. As a high-resolution spaceborne SAR satellite, TerraSAR-X images were used as test data. By situating several well-distributed trihedral corner reflectors within sites and imaging these sites using TerraSAR-X＇ s stripmap （SM） mode, the modeling quality of the delivered stereo pair was validated, and the 3D mapping potential was also assessed. The experimental results indicated that the proposed RPC-based adjustment model could well model the geometric distortion of the TerraSAR-X stereo pair. Furthermore, under the four corner ground control points condition, the proposed model could achieve high accuracy with 0. 166 m in the plan, 0. 241 m in the height.