The RFM has been considered as a generic sensor model. Compared to the widely used polynomial models,
RFM is essentially a more generic and expressive form. Utilizing the RFM to replace physical sensor models in
photogrammetric mapping is becoming a standard way for economical and fast mapping from high-resolution imagery.
This model uses the Rational Polynomial Coefficients (RPCs) supplied with the images, since IKONOS precise sensor
and orbit parameters are not released by the satellite company. This paper presents three mathematical models for
performance enhancement of RFM using IKONOS stereo satellite images, namely: 1) Bias-corrected image space; 2)
Bias-corrected RPCs; and 3) Bias-corrected ground space. The three models were tested and compared with the wellknown
3D-Affine and Direct Linear Transformation (DLT) models. The Least Squares Method (LSM) was applied to
implement the different mathematical setups for estimating the correction parameters. Attained results show that the
accuracies of the five models are slightly variant. With five GCPs, an accuracy of 0.8 m in X, 1.2 m in Y, and 1.3 m in
height is achieved using the bias corrected image space and an accuracy of 0.9 m in X, 1 m in Y, and 1.6 m in height is
reached using the bias corrected RPCs. On the other hand, the results indicate the effectiveness of 3D-Affine and DLT
models especially when the RPCs and/or commercial software packages are not available for users. |
