一种多线形车道线检测算法

A multiple lane detection algorithm

文章提出了一种多线形车道线检测算法。该算法在灰度图上利用图像的梯度信息及车道线的宽度特征提取车道线边缘点,分区域对车道线边缘点进行Hough变换来拟合直线,根据分区中的直线确定控制点并使用三次均匀B样条曲线拟合车道线。在拟合车道线时,利用车道的消失线反求虚线车道的车道标识线,使算法能够对多种线形的车道线进行有效检测;在灰度图上提取目标点,增强了算法的鲁棒性。经过多种场景的试验验证,该算法能够准确、迅速地检测多种线形的车道线。

This paper presents a multiple lane detection algorithm.The gradient information of the image and the width of the lane are used to extract the edge points of the lane on the grayscale image.Hough transform is used to fit the lane edge points in the sub-areas.The control points are determined according to the lane lines in the sub-areas and fits the lane lines using the cubic uniform B-spline.When fitting the lane lines,the use of the vanishing line of lane to get the lane lines of the dotted lane enables the algorithm to effectively detect multiple lane;extracting the edge points on the grayscale image enhances the algorithm robustness.The experiments of multi-scenario show that the algorithm can accurately and quickly detect multiple lane.