A robust scene matching method for mountainous regions with illumination variation

This paper presents a robust scene matching navigation method for mountainous regions with illumination variation. In order to solve the problem of the inconsistency between the reference map and the real map caused by different lighting conditions in mountainous areas, the reference image is obtained by fusing the hillshade map, generated from elevation data using illumination model, and the valley saliency map, generated from water flow data collected by Digital Elevation Model (DEM). To avoid the mismatch caused by the difference of edge between the reference and the real maps, a Hausdorff distance edge matching algorithm based on morphological constraints is proposed. The experimental results of LANDSAT image and ASTERDEM data show that the proposed method has high accuracy and good robustness.