A generic compensation method for dynamic systematic errors in the geolocation of linear pushbroom satellite imagery

Abstract

In Earth observation and planetary exploration missions, high-precision geolocation through remote sensing is crucial. However, dynamic systematic errors in satellite imagery and ancillary flight data, arising from sensor limitation and measurement technology, challenge traditional correction methods such as orientation image model and bias compensation model. This study proposes a piecewise bias matrix compensation method to improve the geolocation accuracy of linear pushbroom cameras by effectively eliminating dynamic systematic errors. The method focuses on detect orientation point exposure time utilizing the distribution characteristics of the reprojection residuals caused by dynamic system errors. This is accomplished through a stepwise strategy involving residual point cloud thinning and iterative end-point fitting algorithms, which automatically divide the sub-compensation model. Furthermore, considering the stability of the satellite’s flight attitude, a Lagrange interpolation method is introduced in the piecewise bias matrix compensation model, enhancing the internal consistency and absolute accuracy in the geometric positioning process of the optical linear pushbroom satellite. A comparison of the initial camera positioning results in the experimental dataset revealed that the reprojected errors of conjugate points in the MEX-HRSC can now achieve sub-pixel accuracy, with a maximum root mean square error of 0.54 pixels; the reprojected errors of GCPs in the XX-18 can now achieve sub-pixel accuracy, with a maximum root mean square error of 0.80 pixels. Additionally, building upon the elimination of dynamic systematic errors, this study constructs high spatial resolution DTMs of the Martian surface using high-resolution stereo imagery from the MEX-HRSC. These techniques and methods result in the intersection of corresponding rays in stereo pairs, which is crucial for subsequent 3D reconstruction.