Accuracy Assessment of Geometric Correction of High Resolution Satellite Images Based on Control Information

The geometric correction of high resolution satellite images (HRSIs) using proper mathematical models is an important step in 3D spatial information extraction process. To do it, some control information such as points, lines or areas should be extracted at first. Then, the correspondence features are determined and used directly to estimate the unknown parameters of a mathematical model. Since, the accuracy of geometric correction of satellite images directly influence the accuracy of georeferencing procedure, in this paper, the issue of solving the unknown parameters of a rational function model (RFM) based on linear features as well as their combination with point features as control information is studied. In addition, the impact of each control information to reduce or omit the systematic errors are also studied. The obtained results show that the accuracy of rational models based on control lines is very low and about 3 pixels. Additionally, the residual vector diagrams demonstrate the existence of enormous amounts of systematic errors. The reason in addition to the accuracy and distribution of control lines could be due to the characteristics of linear features used as control lines. In other hand, the integration of control lines and points to estimate the unknown parameters of the model increases the accuracy as well as reduces the systematic errors significantly. Therefore, the results show the high potential of the integration of point and linear features to improve the accuracy as well as reducing the systematic errors.