Geometric processing of QuickBird stereo imagery with high building data in megapolis

Much research work has been done concerning the geometric accuracy for High-Resolution Satellite Imagery (HRSI), the conclusion regarding this respect in photogrammetric society is that Ground Control Points (GCPs) are necessary for the accurate geo-positioning to get meter or sub-meter positioning accuracy. But most existing publications are about the horizontal positioning determination rather than vertical direction. This paper aims to evaluate the geo-positioning capability in vertical direction for HRSI and analyze the influence of GCPs vertical distribution to that of the final vertical accuracy. The study is performed in Shanghai, China, the typical metropolis, using a pair of across-track QuickBird stereo imagery, within a relatively small but representative area filled with very high buildings to provide height difference for GCPs and Independent Check Points (ICPs). A new accuracy evaluation scheme in vertical direction is put forward and employed in the assessment. Processed with the Rational Function Model (RFM) and its accuracy improvement models in image space, some relations are found between the GCPs height and obtainable ICPs vertical accuracy in different height ranges, discussions and conclusions are presented in the end.