New software for RPC use and generation

In the last years the commercial availability of high resolution satellite data has opened a large number of new opportunities for the operative use of the Remote Sensing data. Nowadays it is possible to carry out many applications that in the recent past were exclusive of the airborne and on site surveys. A key issue for mapping information from space is the geometrical processing of satellite data that is mandatory when a full exploitation of satellite data capabilities is required. The remote sensing community usually adopts two different types of geometrical models: the physical sensor models and the generalized sensor models. In the first ones involved parameter has a physical meaning; besides, they require the knowledge of the specific satellite and orbit characteristics. On the contrary the generic models are based on the Rational Polynomial Functions (RPF) and do not need the knowledge of sensor features, allowing to be employed also by noexpert users. Due to these attractive characteristics, many satellite imagery vendors have considered the use of Rational Function Models (RFM) as a standard to supply a re-parametrized form of the sensor model in term of the so-called Rational Polynomial Coefficients (RPC), secretly generated from they physical sensor model. Driven by these motivations, in this research an investigation over the RPF application and Rational Polynomial Coefficients generation was conducted developing and implementing an original algorithm in C++ language, starting from the physical sensor model developed at the Section of Geodesy and Geomatic of “Sapienza”, University of Rome and embedded in the software SISAR (Software per Immagini Satellitari ad Alta Risoluzione). The present research is divided in two main parts. The former concerns with the definition of an algorithm for the Rational Polynomial Coefficients (RPC) application, supplied with QuickBird and IKONOS II imagery. In order to upgrade the results and to eliminate bias errors, a refinement model was developed, estimating some additional parameters on the basis of available Ground Control Points (GCP). The latter considers the definition of an innovative and powerful algorithm for the RPC extraction from the rigorous sensor model of two satellites – QuickBird and EROS A (the last does not provide RPC in the images data files) - utilizing a terrain independent approach. All methods were tested on images with different features; for each image the model intrinsic precision and the image accuracy were assessed computing Root Mean Square Error (RMSE) of Ground Control Points (GCP) and Check Points (CP) residuals. The SISAR results were finally compared with the results obtained by two commercial software (OrthoEngine v. 10.0, ERDAS v 9.0).