Parameterschätzung des RPC-Models für Satellitenbilder mittels Bodenpasspunkte

Abstract

Parameterschätzung des RPC-Models für Satellitenbilder mittels Bodenpasspunkte

The importance of high-resolution satellite imagery has been growing steadily in recent decades. One application is the determination of ground coordinates from satellite images. Therefor the orientation parameters and the sensor model of these images are necessary. One sensor model is using rational polynomial coefficients (RPCs) and gained in importance in recent years. This RPC-Model will be analyzed in this paper. The calculation of the coefficients of the RPC-Model are terrain independent. So a verification with ground control points is possibly. Now different models can be used to improve the accuracy of the original RPC-Model. One model is an additional translation to reduced the bias shifts. Another model is using an affine transformation to compensate shifts, drifts and rotations. These two models extend the RPC-Model and increase the number of parameters. Also a new way to improve the orientation of the satellite images will be discussed in this paper: the direct estimation of the RPCs with ground control points. First some diverent sensor models and the RPC-Model gets introduced. Then a mathematic method will be presented to estimate object coordinates with this model. Building on this the translation, the affine transformation and the direct estimation of the RPCs will extend the model. These models are then examined with four Pleiades satellite images and compared to each other and the advantages of direct estimation of the RPCs are shown.