基于三维激光雷达的道路边界提取方法

Extraction method of road boundary based on three-dimensional LiDAR

为精准提取道路边界,为道路管理及智能交通规划提供基础,本文提出基于三维(3D)激光雷达的道路边界提取方法,具体实现步骤为:对于采集的激光点云图像,通过判断图中某一点云与其邻近点的高差平均值去除点云图像中的噪声;结合激光射线特征与平面模型拟合的方法,分别从粗分割与细分割两个阶段从去噪后的点云中分割出地面点云;将地面点云中邻近点高度差、平滑度以及邻域点角度当作道路边界提取特征并构建多特征融合策略,判断点云中的点是否为道路边界点;针对道路边界,构建二次抛物线模型,采用最小二乘法拟合模型系数完成道路边界提取。为了验证本文方法的有效性,利用本文方法对某段道路进行边界提取实验。实验结果显示:该方法可以有效提取道路边界,且精准度高,实验结果表明:本文方法在道路边界提取问题中有着较好的表现,具有较强的稳健性和适用性,可以满足智能交通规划以及无人驾驶车辆自主导航的需要。

To accurately extract road boundaries and provide a foundation for road management and intelligent transportation planning,this paper proposed an extraction method for road boundaries based on three-dimensional(3D)light detection and ranging(LiDAR).The specific implementation steps were:In terms of the collected laser point cloud image,the noise in the point cloud image was removed by judging the average height difference between a point cloud in the image and its neighboring points.By combining laser ray features with plane model fitting methods,ground point clouds were segmented from the denoised point cloud in two stages,including coarse segmentation and fine segmentation.The height difference,smoothness,and angle of neighboring points in the ground point cloud were taken as road boundaries to extract features and construct a multi-feature fusion strategy,which could determine whether the points in the point cloud are road boundary points.A quadratic parabolic model for road boundaries was constructed,and the boundary was extracted by fitting the model coefficients with the least squares method.To verify the effectiveness of the proposed method,boundary extraction experiments were conducted on a certain section of the road by employing the proposed method.The experimental results indicate that this method can effectively extract road boundaries with high accuracy,and this method shows good performance in problems of road boundary extraction,strong robustness and applicability.This method can meet the needs of intelligent transportation planning and autonomous navigation of unmanned vehicles.