Parametric Refinement of the Rational Function Model for Satellite Remote Sensing Imagery

To date, the rational function model (RFM) is widely used in geometric calibration and orthorectification of satellite remote sensing imagery. Influenced by the measurement accuracy of the satellite altitude and attitude, parameters of the RFM (also call as RPC) are required to be corrected. Traditionally, grid points are generated as virtual control information and RPC is refined by fitting all points, which will introduce extra fitting error and consuming more time. Hence, this paper introduces a parametric refinement method is introduced to reproduce RPC. Firstly, a weighted NCC method is introduced for automatic extraction of GCPs. By combining the bias compensation model and RFM based on GCPs, we can derive the relationship between the refined RPC and original RPC. Finally, parametric expressions of the refined RPC can be obtained. Multiple sources Optical and SAR satellite remote sensing imagery covering the Mount Song area are experimented to evaluate the performance of our proposed method. Compared with traditional methods, the proposed method gives a fast and efficient way for the refinement of RPC.