Improved Lane Line Detection Algorithms Based on Incomplete Line Fitting

QingYu Ren, Bingrui Zhao, TingShuo Jiang, Weize Gao

Abstract: Lane detection is a crucial component of autonomous driving, as it enables vehicles to accurately perceive their environment and safely navigate roads. However, existing lane detection algorithms often struggle to balance real-time processing and robustness. In this paper, we propose an improved sliding window lane detection algorithm that overcomes this challenge. Our algorithm begins by transforming the image into a bird’s-eye view using inverse perspective transformation and then generates a binary map using the Sobel operator. The algorithm then dynamically selects between quadratic curves and straight lines to fit the vacant portions of the left and right lane lines within the current window and determines the position of the next sliding window. Finally, a quadratic curve or straight line is dynamically selected to fit the entire lane trajectory using the least square method. Our testing results show that our algorithm can detect lane lines quickly while maintaining high robustness and accuracy. This algorithm has the potential to be a valuable contribution to the field of computer vision and image processing for autonomous driving.

External IDs:dblp:conf/icic/RenZJG23