Orthorectifying Archive Aerial Photos with Rational Polynomial Camera Model

Abstract

In this paper we present a workflow, how a set of metadata can be calculated for a raw aerial image, that enables end-users to simply orthorectify the aerial image, using end- user’s own digital elevation model. The set of metadata mentioned here are the coefficients of the Rational Polynomial Camera (RPC) model. The Rational Polynomial Camera model is an extension of the rigorous collinearity equations (used for frame camera orthocorrection) with higher power terms of the object space coordinates. We used the fiducial points of the image to estimate the interior orientation parameters and we used Ground Control Points (GCPs) to estimate exterior orientation parameters of the photo. As the GCPs were measured in map projection system, and the RPC standard requires ellipsoidal coordinates as object space coordinates, we also present the methodology, how we calculated the local transformation coefficients between these two systems. All these efforts resulted in the calculation of the RPC coefficients with closed form equations. We generated the orthophoto using the RPC model, and the accuracy of the orthophoto equals to the accuracy obtained using collinearity equations. Besides the collinearity model representing an ideal camera, we also calculated RPC coefficients additionally considering the radial lens distortion. From this set of RPC coefficients we also generated the orthophoto. Comparing these two results, we found, that considering radial distortions did not significantly improve the accuracy of the generated orthophoto. To make our results repeatable we also present the calculated RPC coefficients in the standard form, enabling the reader to generate the orthophoto.