Pros and Cons of the Orientation of Very High Resolution Optical Space Images

Very high resolution optical space images are coming more and more into competition to classical aerial images making them more important for mapping projects. The orientation process of the dominating CCD-line scanner images taken from space is quite different from the orientation of perspective photos. Different image products like nearly original images and images projected to a plane with constant height are available, this requires corresponding orientation solutions. All optical satellites are equipped with direct sensor orientation based on a positioning system, gyros and star sensors, so they can estimate the orientation parameters. For IKONOS, QuickBird and OrbView-3 this can be made without control points with a standard deviation of the ground coordinates in the range of 10m and better. The orientation information is available as view direction from the scene centre or start of scene or as ephemeris and as rational polynomial coefficients (RPCs). This allows the determination of the scene orientation by geometric reconstruction or bias corrected RPCs using control points. In addition approximations like 3D-affine transformation, direct linear transformation and terrain related RPCs are in use. The accuracy of the methods is compared in relation to the number and distribution of the control points and the limitations of the approximations are investigated. The approximations of the orientations by 3D affine transformation and direct linear transformation can only handle images projected to a plane with constant height like IKONOS Geo and QuickBird OR Standard. For original images they are not leading to acceptable accuracy. In any case they require more and three-dimensional well distributed control points.