Evaluation of area-based image matching applied to DTM generation with HiRISE images

In the generation of Digital Terrain Models (DTMs) from space images area-based image matching techniques are used in most cases; the image matching strategy, though, is to be tailored to the specific mission characteristics. In this paper, a series of tests have been performed with real and synthetic space images to study the performance of our implementation of the Least Squares Matching algorithm. To this aim, patches extracted from a pair of Hirise images were processed with our DTM generation software Dense Matcher (DM) as well as with Ames Stereo Pipeline (ASP) and the results compared to a reference DTM; moreover, synthetic images, generated from the reference DTM and overlaid with artificial texture ideal for image matching, were processed as well, to discriminate the influence of texture and surface topography on the DTM accuracy. With real images, both software show RMSE ranging from 15 to 38 cm; with synthetic images, processed only with DM, RMSE ranges from 12 to 18 cm.