Geopositioning Accuracy Analysis Of Physical Model And RPC Model In Q.B. High Resolution Imagery

The high resolution satellite imagery (HRSI) has been widely used in recent years because it provides the capabilities of high precision geo-positioning and large scale mapping. So HRSI is being widely used in many fields. At present many sensor models have been presented to describe the geometrical relationships between object space and image space. In conventional field—photography, physical model that reaches maturity and has the feature of high resolution is mostly used. However physical model is of limitation in application because of its sophisticated imaging geometry, complicate sensor physical structure, and self-correlation of model parameters. For the purpose of technology protection, some image vendors only use Rational Function Model (RFM), and a set of Rational Polynomial Coefficients (RPC) is given as parameters of the RFM.Besides, some other imaging models, such as affine model, polynomial function and direct linear translation model, are also used in geopositioning. Hence it becomes a heat topic to study the geopositioning accuracy when these sense models are used in HRSI.Firstly, a review is performed about the HRSI geo-positioning model development, and the geo-positioning principle of physical model and RPC are discussed for a comparison of application. Secondly, by making use of the obtained QuickBird data and the original model information provided by Digital Globe Company, an initial physical sensor model is established and a bundle adjustment method is used to improve the geo-positioning accuracy. Thirdly, in order to better fit Shanghai area, the rational polynomial coefficients are recomputed through the iterative least-squares method based on terrain dependent and terrain independent methods, respectively, combined with enough ground control points (GCPs). On the other hand, because RPC is the parameterization of physical model, the error of interior and exterior orientation parameters leads to the error of RPC. So for the sake of improving accuracy, paper has established both direct correction model and indirect correction model in RFM, and employs adjustment method with additional GCPs. Finally the feasibility analysis of these models is discussed. At last, the influence on RPC result from GCPs is then analyzed with respect to the amount, quality and distribution of GCPs. The accuracy comparison is also given for the purpose of improving mapping scale from point of applicability, merits, faults and limitation. Conclusion in this paper can be considered as foundation of HRSI data processing, so the customers can choose appropriate image geopositiong method according to their own data feature to get a good accuracy. And the results in this paper can also act as quality control during data pro-processing.