Numerical evaluation and application-oriented analysis for forward and inverse rational function models of terrain-independent case in satellite imagery
Terrain-independent Rational Polynomial Coefficients (RPCs) are considered as most important part of the optical satellite and aerial imagery data processing especially those ones with high resolution since the proposed RPCs by the aerospace companies have some limitations in particular for using directly by the geoscientists in environmental studies and other Earth observation applications. While the inverse RPCs have more advantageous rather than direct ones, in this study, a new approach is presented in order to provide the inverse RPCs from direct ones and also to satisfy satellite imagery products users. In order to do this, first a spatial 3D-cubic is going to be fitted to the study area approximately including necessary altimetry layers numbers. Next, a range of virtual control points are being created in those altimetry layers randomly and then these points are going to be shifted to the image space by means of given direct RPCs. Hence, the inverse RPCs computes from the direct ones by space resection technique. Finally, the ground coordinates for the corresponding points have derived from different space intersection methodologies, direct RPCs and also inverse ones. Moreover, comparative tests have been developed to assess the effects of different altimetry layers numbers and also the number of virtual control points on the quality of derived inverse RPCs. It is demonstrated here that the precision of derived RPCs are increasing as much as the number of altimetry layers and control points increase. The proposed methodology, computations, data processing and results evaluation are discussed in details.