A Robust Lane Detection Method Based on Vanishing Point Estimation

Abstract

Although lane detection technology has been developed for decades, however, many challenging issues remain unresolved. In this paper, we propose a robust vanishing point-based lane detection method. Due to the perspective effect, the two parallel straight lines in 3-D space intersect at the 2-D plane. For the lane line extraction, we use lane shape features to extract lane lines. Firstly, we use LSD (line segment detectors) to extract the straight line segments in each frame of image. Secondly, we use a direction priority search method to remove the most of the interference information. This algorithm makes reference to the directional and shape features of lane lines in 2-D images. Finally, using the remaining straight line segments after filtering out the noise straight line to calculate the vanishing point of the lane line, we use a score function to remove non-candidate lane markings, the score function is constructed using the shape features, direction of the line segments. The experimental results show that the validity and robustness of our new algorithm under complex structured road scenes.