Satellite SAR geocoding with refined RPC model
Satellite SAR geocoding with refined RPC model
复制标题
使用精炼的 RPC 模型进行卫星 SAR 地理编码
DOI：
10.1016/j.isprsjprs.2012.02.004
复制
发表时间：
2012-04-01
影响因子：
12.7
通讯作者：
Liao, Mingsheng
中科院分区：
文献类型：
--
作者：
Zhang, Lu;Balz, Timo;Liao, Mingsheng
Recent studies have proved that the Rational Polynomial Camera (RPC) model is able to act as a reliable replacement of the rigorous Range-Doppler (RD) model for the geometric processing of satellite SAR datasets. But its capability in absolute geolocation of SAR images has not been evaluated quantitatively. Therefore, in this article the problems of error analysis and refinement of SAR RPC model are primarily investigated to improve the absolute accuracy of SAR geolocation. Range propagation delay and azimuth timing error are identified as two major error sources for SAR geolocation. An approach based on SAR image simulation and real-to-simulated image matching is developed to estimate and correct these two errors. Afterwards a refined RPC model can be built from the error-corrected RD model and then used in satellite SAR geocoding. Three experiments with different settings are designed and conducted to comprehensively evaluate the accuracies of SAR geolocation with both ordinary and refined RPC models. All the experimental results demonstrate that with RPC model refinement the absolute location accuracies of geocoded SAR images can be improved significantly, particularly in Easting direction. In another experiment the computation efficiencies of SAR geocoding with both RD and RPC models are compared quantitatively. The results show that by using the RPC model such efficiency can be remarkably improved by at least 16 times. In addition the problem of DEM data selection for SAR image simulation in RPC model refinement is studied by a comparative experiment. The results reveal that the best choice should be using the proper DEM datasets of spatial resolution comparable to that of the SAR images. (C) 2012 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS) Published by Elsevier B.V. All rights reserved.
