基于RRF的车道理解算法

Algorithm of Lane Recognition Based on RRF Method

基于视觉的道路图像提供了车辆运行局部环境的丰富信息。本文提出了一种新颖的适用于自主驾驶系统的车道线检测与跟踪算法。它采用了广义曲线的车道线参数模型,通过计算机仿真实验,将图像的预处理算法和基于坐标变换原理的拾取算法相结合(即RRF算法),在感兴趣区域中成功提取道路边缘。最后,在2维计算机图像坐标系中重建车道线曲线,为以后的数据提取打下了基础。仿真结果表明,提出的车道线检测与跟踪算法具有良好的鲁棒性和实时性。

Large amounts of information can be obtained from road images captured by computer vision.This paper presents a novel lane detection and tracking algorithm for automatic drive system.Thealgorithm utilizes the generalized curve lane parameter model.It has combined image preprocessingalgorithm with lane extraction algorithm based on the coordinate transformation,then,it had been donesuccessfully to extract the lane marks in ROI(named RRF).Furthermore,lane curve was rebuilt in a2-Dcomputer coordinate system,which has lay down the foundation for the future research.In the end,thesimulation results validate the correctness and security of this algorithm.