论文部分内容阅读
针对结构化道路上存在非车道线标记干扰的情况,提出一种基于成像模型的线扫描车道线检测及跟踪方法。检测算法中首先对路面图像进行形态学高帽变换预处理,然后建立前方道路图像的成像模型,将图像坐标系中车道参数和世界坐标系中实际车道参数对应,对图像进行初扫描,利用边缘贡献函数及RANSAC算法选取最确定线后,以此线为标准进行二次扫描,得到边缘点后统计边缘贡献函数局部最大值并拟合成直线车道线。跟踪算法中运用Kalman滤波器预测车道线区域,并提取符合标准的控制点拟合成模型为B样条的车道线。试验结果表明:该方法能够快速准确地在复杂环境中提取多个车道线,尤其对存在非车道线道路标记干扰的情况有显著效果。
In view of the existence of non-lane marking interferences on the structural road, a scanning line detection and tracking method based on imaging model is proposed. In the detection algorithm, the image of the road surface is first transformed into a morphological top hat shape, and then the imaging model of the road image in the front is established. The lane parameters in the image coordinate system are corresponding to the actual lane parameters in the world coordinate system. After the contribution function and the RANSAC algorithm select the most deterministic line, this line is used as the standard for the second time scan. The local maximum of the edge contribution function is obtained after the edge point and the straight line is fitted. The Kalman filter is used in the tracking algorithm to predict the area of the lane line, and the standard control points are extracted to fit the lane line with the B-spline. The experimental results show that this method can extract multiple lane lines in complex environment quickly and accurately, especially for the case of lane mark interference in non-lane line.