Internal approach for the geometric accuracy evaluation of some orthorectification models applied to quickbird images

The high and very high space resolution sensors (VHR) outline new cartographic challenges on cartographic design and end-product quality control. In order to integrate VHR images into the production of image cartography at medium and large scales, this study concerns the geometric mechanism of VHR images in the orthorectification process working with QuickBird Basic Imagery panchromatic satellite images. The subject was to carry out some control methods to evaluate the geometric accuracy of potential image maps derived of these images. A QuickBird image, acquired over low-relief agricultural zone located 25 Km from the city of Salamanca (Spain), is processed using different geometric modelling and the accuracy of each modelling is evaluated by different procedures. A new approach is presented to check the internal shifts between the different methods. A kind of fiducial marks were added into the scene to create a ‘reseau grid’. It lets to control the deformation of several good distributed points called Tie Internal Points (TIP). The control grid gets a good advantage because it controls how the different methods fix in whole warped scenes without knowing terrain coordinates for all of the area. The process is validated by merging TIP, Check Points (CPs), vectorial maps and linear trajectories from GPS devices. TIP show the relative warping for areas without external control. It became a good way to estimate the geometric accuracy and it is coherent with the other method which use terrain coordinates.