Towards A More Rigorous Boresight Calibration 1

This work focuses on the stochastic properties of boresight determi nation between a strapdown IMU and a framebased imaging sensor. The core of the stochastic model is a rigorous error propagation of the estimated input accuracies and their correlations. This subsequently yields mor e realistic estimates of boresight accuracy for every image event. The nature of INS/GPS integration makes the residual orientation errors be strongly correlated within a flight line while little horizontal acceleration is sensed. Thi s information is partially carried within the Kalman filter used for the data integration; however, it is rarely m ade accessible to the user. As an alternative, we propose estimating the temporal correlations by means of a simp lified function that is derived from the flying profile and the inertial system used. Numerical examples are demonstrat ed for the case of high-resolution digital imagery, where the application of the proposed method allows reaching sub-pixe l accuracy in direct georeferencing (e.g. forward intersection without an adjustment).