Sub-metre Geopositioning with Ikonos Geo Imagery

The Ikonos one-metre earth observation satellite system offers the photogrammetric and remote sensing communities a significant new means for geospatial information collection, with the potential of pixel-level geopositioning precision. This paper addresses the subject of the metric accuracy potential of Ikonos Geo imagery for both 2D and 3D geopositioning. Photogrammetric approaches to metric accuracy evaluation of line scanner imagery typically involve application of collinearity-based restitution models which require sensor calibration data and possibly also information related to exterior orientation. With Ikonos data, however, both the camera model and precise satellite ephemeris data are withheld from the user of Geo imagery, leaving only two practical alternatives for multi-image triangulation. The first is use of rational functions and the second involves the employment of sensor orientation models which do not rely on knowledge of the camera model, but do require ground control. Two such candidate approaches are the affine projection model and an extended direct linear transformation (DLT). A test field of precisely measured ground control points within the city of Melbourne has been imaged with 3-fold Ikonos 1m Geo coverage comprising a stereopair and a near nadir-looking image. In order to quantify the metric potential of this imagery, both 2D and 3D geopositioning have been investigated, the former primarily to verify sensor integrity (linearity) and the latter to ascertain accuracy in the presence of a modest provision of ground control. In both instances sub-pixel accuracy was attained, with planimetric positioning to 0.3m and heighting reaching 0.7m accuracy. The paper describes the Melbourne testfield, discusses the 2D and 3D geopositioning approaches adopted, and reports on the geometric accuracy obtained with the different sensor orientation models.