Rational function model based geo-positioning accuracy evaluation and improvement of QuickBird Imagery of Shanghai, China

Rational Function Model (RFM) is introduced and RFM-based ground positioning algorithm is derived. Experiments of geo-positioning are conducted based on the Rational Polynomial Coefficients (RPCs) provided in the image support data using QuickBird separate-orbit stereo imagery pair in Shanghai district, China. Positioning errors are then analyzed. According to the error varying characteristics, several compensation models are introduced and used to refine the positioning accuracy. The experimental results show that with only one control point (CP), the positioning accuracy can be largely improved from 23m to 3m using Offset Model. The Offset plus X-scale Model with 2 well distributed CPs can improve the accuracy further to sub meter in planimetry and 1-2m in elevation. Overall, the Affine Model works best for the experimental area. The accuracy turns to be steady and closes towards 0.6m in planimetry and 1m in elevation when more than 4 well distributed control points are available using Affine Model either in image space or in object space.