AN ITERATIVE PIXEL-LEVEL IMAGE MATCHING METHOD FOR MARS MAPPING USING APPROXIMATE ORTHOPHOTOS

Abstract. Mars mapping is essential to the scientific research of the red planet. The special terrain characteristics of Martian surface can be used to develop the targeted image matching method. In this paper, in order to generate high resolution Mars DEM, a pixel-level image matching method for Mars orbital pushbroom images is proposed. The main strategies of our method include: (1) image matching on approximate orthophotos; (2) estimating approximate value of conjugate points by using ground point coordinates of orthophotos; (3) hierarchical image matching; (4) generating DEM and approximate orthophotos at each pyramid level; (5) fast transformation from ground points to image points for pushbroom images. The derived DEM at each pyramid level is used as reference data for the generation of approximate orthophotos at the next pyramid level. With iterative processing, the generated DEM becomes more and more accurate and a very small search window is precise enough for the determination of conjugate points. The images acquired by High Resolution Stereo Camera (HRSC) on European Mars Express were used to verify our method’s feasibility. Experiment results demonstrate that accurate DEM data can be derived with an acceptable time cost by pixel-level image matching.