Availability of detailed, high accuracy digital maps is crucial for many driver assistance applications. In this paper, a novel approach for automated map generation from aerial images is presented. Automatically generated maps present an alternative for applications that are not critical for vehicle safety such as accurate map based localization for route guidance. Aerial images as a source for automated map generation are available worldwide and updated frequently, which makes short map update cycles practicable. The paper describes the generation of landmark maps from aerial images, consisting of a per-pixel image classification and the computation of a geometric representation of the landmarks. A special image descriptor for aerial image classification is introduced, which helps to reduce the required amount of manual post-processing significantly. We further demonstrate a method for lane boundary estimation based on the extracted landmarks. Results of the map generation algorithm are presented for an inner-city intersection scene.
Automated map generation from aerial images for precise vehicle localization
2010-09-01
1106968 byte
Conference paper
Electronic Resource
English
Precise docking control in unmanned aircraft vehicle automated aerial refueling
British Library Online Contents | 2011
|Vehicle Ego-Localization by Matching In-Vehicle Camera Images to an Aerial Image
British Library Conference Proceedings | 2011
|