DocumentCode
2220707
Title
Single camera lane detection and tracking
Author
Schreiber, David ; Alefs, Bram ; Clabian, Markus
Author_Institution
Advanced Comput. Vision GmbH, Vienna, Austria
fYear
2005
fDate
13-15 Sept. 2005
Firstpage
302
Lastpage
307
Abstract
In this paper we present a method to detect and track straight lane boundaries, using a forward-looking single camera mounted on-board. The method proved to be robust, even under varying lighting conditions, and in the presence of heavy shadowing cast by vegetation, vehicles, bridges, etc. Moreover, false positive hardly occur. The lane markings are being continuously detected even when the vehicle is performing maneuvers such as excursion or lane change. The performance is achieved by a novel combination of methods, namely, by first finding vanishing point candidates for the lane boundaries, and later by the careful selection of line segments given a vanishing point, and finally by using a smart heuristics to select the whole set of four lane boundaries. Suppression of false positives is further achieved by tracking the vanishing point, and by constraining the lane width based on recent history.
Keywords
image sensors; object detection; road vehicles; traffic engineering computing; false positives; lane markings; lane tracking; single camera lane detection; single camera mounted on-board; vanishing point; vision based intelligent vehicles; Cameras; Infrared detectors; Infrared imaging; Intelligent transportation systems; Intelligent vehicles; Pixel; Roads; Robustness; Vehicle detection; Voting;
fLanguage
English
Publisher
ieee
Conference_Titel
Intelligent Transportation Systems, 2005. Proceedings. 2005 IEEE
Print_ISBN
0-7803-9215-9
Type
conf
DOI
10.1109/ITSC.2005.1520207
Filename
1520207
Link To Document