Title :
Sharp curve lane boundaries projective model and detection
Author :
Yong Chen ; Mingyi He
Author_Institution :
Shaanxi Prov. Key Lab. of Inf. Acquisition & Process., Northwestern Polytech. Univ., Xi´an, China
Abstract :
An effective lane boundaries projective model (LBPM) and improved detection method in the images captured with a vehicle-mounted monocular camera in complex environments, especially for sharp circular curve lane, is proposed in this paper. Firstly, a lane boundaries projective model is deduced. This lane model can not only express straight-line lane boundaries, but also describe the actual sharp circular curve lane boundaries very well. Secondly, the lane posterior probability function is derived by employing the lane model, the gradient direction feature, the lane likelihood function, and the lane prior information. And then the lane maximum posteriori probability is found out by using the improved particle swarm optimization algorithm. Further the lane boundaries is positioned, and the lane geometric structure, such as the lane left and right boundaries curve radiuses, can be calculated accurately through the lane model. The experimental results show that the proposed lane boundaries projective model and the improved detection method are more effective and accurate for sharp curve lane detection.
Keywords :
cameras; edge detection; maximum likelihood detection; particle swarm optimisation; probability; roads; traffic engineering computing; LBPM; actual sharp circular curve lane boundaries; complex environments; gradient direction feature; image capturing; improved detection method; improved particle swarm optimization algorithm; lane geometric structure; lane left boundaries curve radiuses; lane likelihood function; lane maximum posteriori probability; lane model; lane posterior probability function; lane prior information; lane right boundaries curve radiuses; sharp circular curve lane detection; sharp curve lane boundaries projective model; straight-line lane boundaries; vehicle-mounted monocular camera; Cameras; Equations; Feature extraction; Mathematical model; Roads; Shape; Vehicles; lane detection; lane likelihood function; lane model; lane posterior probability;
Conference_Titel :
Industrial Informatics (INDIN), 2012 10th IEEE International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4673-0312-5
DOI :
10.1109/INDIN.2012.6301186