Robust lane localization using multiple cues on the road

Abstract

This paper presents a method for estimating the current lane number in which the vehicle is traveling. An important component of visual automobile safety systems rely on knowledge of the lane number to know how far the vehicle is from the exit lane. The method uses taking the Inverse Perspective Map of an image to get a top view of the road and then detects multiple lanes using Hough transform line detection. Then the algorithm classifies the lines as solid or broken and also detects if the lane is branching or split using connected component analysis. This knowledge of the line type and whether the lane is splitting is the basis to identify if the vehicle is in the exit lane.

8 Figures and Tables

Cite this paper

@article{Pavuluri2013RobustLL, title={Robust lane localization using multiple cues on the road}, author={S. Pavuluri and Kikuo Fujimura and Sally Wood}, journal={2013 IEEE Digital Signal Processing and Signal Processing Education Meeting (DSP/SPE)}, year={2013}, pages={141-146} }