Title :
Self-calibrated multiple-lane detection system
Author :
Jiang, Yan ; Gao, Feng ; Xu, Guoyan
Author_Institution :
Sch. of Transp. Sci. & Eng., Beihang Univ., Beijing, China
Abstract :
We presented a multiple-lane detection system in this paper. To cope with both straight and curved road, the system recognized the road type based on the vehicle´s GPS position and digital map data, and designated different schemes for detection. Unlike the known systems that have treated lane marking as a single entity, this system focused on the inner edge of the stripe to locate a vehicle in a lane accurately. The camera´s installation parameters, especially the pitch angle, were estimated using the real-time detection results to compensate for changes in the vehicle´s pose. Using an “estimate-detect” style, adjacent-lane markings were detected even when they were partly blocked by other vehicles.
Keywords :
Automotive engineering; Calibration; Curve fitting; Feature extraction; Global Positioning System; Image edge detection; Roads; Safety; Transportation; Vehicle detection; Multiple lane detection; calibration of pitch angle; inverse perspective mapping; lateral position determination;
Conference_Titel :
Position Location and Navigation Symposium (PLANS), 2010 IEEE/ION
Conference_Location :
Indian Wells, CA, USA
Print_ISBN :
978-1-4244-5036-7
Electronic_ISBN :
2153-358X
DOI :
10.1109/PLANS.2010.5507334