We present a robust and fast method for on-line creation of local maps based on stereo vision for vehicles operating in off-road environments. A 3D vision sensor system with a high accuracy even at long ranges and wide field of view is used as the only sensor input. Due to the hierarchical mode of operation, dense stereo matching on image resolution 1200 x 525 and a disparity range of 280 becomes feasible at 10 fps. Beside of an achieved speedup factor of 6.47, a significant increase in the density of resulting disparity maps on real-world scenes has been achieved. Multiple captured views are aligned and integrated into a probabilistic elevation map suited for modeling dynamic environments. Efficient computations in the u-disparity-space and a stereo sensor model are at the core of the iterative update process. © 2012 Springer-Verlag.
CITATION STYLE
Kadiofsky, T., Weichselbaum, J., & Zinner, C. (2012). Off-road terrain mapping based on dense hierarchical real-time stereo vision. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7431 LNCS, pp. 404–415). https://doi.org/10.1007/978-3-642-33179-4_39
Mendeley helps you to discover research relevant for your work.