城市结构化道路路面点云提取方法研究

Point cloud extraction method of urban structured road pavement

为了有效提取城市结构化道路车载点云场景中道路面,本文提出了一种基于点云法向量分布特征的道路面提取方法。首先,通过一种基于布料模拟的滤波算法(CSF)对原始车载点云进行滤波处理,去除非地面点云干扰;其次,使用主成分分析法对滤波后得到的地面点进行计算,得到各激光点的局部法向量与曲率值;最后,使用改进区域生长算法,以路面点云的法向量相似度为约束条件分割得到路面点云。使用两段不同场景车载点云数据作为试验数据对本文提出方法的有效性进行验证。结果表明,路段1提取道路面点云的检测质量q为95.16%,完整性r为95.82%,准确性p为97.75%;路段2提取道路面点云的检测质量q为94.38%,完整性r为95.63%,准确性p为96.42%,提取结果受道路宽度、形状的影响较小,因此本文提出方法的适用性较强。

In order to effectively extract the road surface in urban structured road vehicle point cloud scene,a road surface extraction method based on point cloud normal vector distribution feature is proposed in this paper.Firstly,a filtering algorithm based on cloth simulation(CSF)is used to filter the original vehicle point cloud to remove the interference of non ground point cloud;Secondly,the principal component analysis method is used to calculate the filtered ground points,and the local normal vector and curvature value of each laser point are obtained;Finally,the improved region growth algorithm is used to segment the road point cloud with the constraint of the normal vector similarity of the road point cloud.The test results show that the detection quality q of road surface point cloud extracted from section 1 is 95.16%,the integrity r is 95.82%and the accuracy p is 97.75%,and the detection quality q of road surface point cloud extracted from section 2 is 94.38%,the integrity r is 95.63%and the accuracy p is 96.42%.The extraction results are less affected by the road width and shape,so the proposed method has wide applicability.