Precision Processing of Hirise Stereo Orbital Images for Topograpic Mapping on Mars

To build HiRISE image network with the best possible precision and accuracy, technical issues involving sensor geometry and absolute positioning need to be resolved. High-frequency random patterns, or “jitter”, in the rotation angles cause disagreements between HiRISE CCDs. To generate seamless topographic products, it is important to achieve coherence in the exterior orientation parameters between multiple CCDs. Also, new images should be registered to the existing global geodetic control network to achieve absolute positioning in global scale. This paper focuses on bundle adjustment based on rigorous sensor modeling of HiRISE imagery and absolute positioning using MOLA control network. For construction of orbital image networks, bundle adjustment system using traditional and inter-CCD tie points is developed to resolve misalignment of exterior orientation parameters between stereo images and CCDs. The bundle adjustment method based on third-order polynomials reduced mean back projection error of traditional tie points from over 3 pixels to less than 0.2 pixel for the test dataset covering Spirit rover landing site. The back projection residual between inter-CCD tie points were also reduced to about half a pixel. For absolute geopositioning without ground-control points, HiRISE DTM is first matched with MOLA MEDGR DTM and then with PEDR profile after bundle adjustment. The absolute positioning result is compared with the Spirit rover landing position visible on the HiRISE image. The displacement is reduced from about 200 m to 65 m, which is successful given the MOLA horizontal accuracy of 100 m.