Road lane modeling based on RANSAC algorithm and hyperbolic model

This paper presents a simple and fast road lane detection and modeling method with high robustness. We use IPM transformation to get a “bird's eye view” of the road and do some image processing to avoid noise, then divide regions for each lane and implement a improved RANSAC algorithm to fit Hyperbolic model defined by us. Our method can successfully detect road lanes in various conditions and achieve reasonable results on KITTI dataset.