基于车载激光扫描点云数据的城市道路面提取方法研究

Research on Extraction Method of Urban Road Surface Based on Vehicle Laser Scanning Point Cloud Data

为了进一步扩展车载激光扫描点云数据的应用领域,针对点云数据的分类尤为重要。本文针对目前道路面提取中存在的问题,提出了一种基于Otsu算法与改进区域生长算法的道路面提取方法。该方法实现道路面精确提取的关键步骤为:首先,使用Otsu算法自适应计算出分割阈值实现非地面点滤除;其次,计算得到空间激光点的法向量与曲率;最后,将法向量相似度作为约束条件,使用改进区域生长算法进行道路面精确提取。使用两段城市道路点云数据进行试验,结果表明本文方法提取道路面结果的准确度CR、完整度CP以及提取质量Q均大于94%,验证了本文方法的有效性与鲁棒性。

In order to further expand the application field of vehicle borne laser scanning point cloud data,the classification of point cloud data is particularly important.Aiming at the problems in road surface extraction,this paper proposes a road surface extraction method based on Otsu algorithm and improved region growing algorithm.The key steps of this method to achieve accurate road surface extraction are as follows:First,use Otsu algorithm to adaptively calculate the segmentation threshold to filter off non-ground points;Secondly,the normal vector and curvature of the space laser point are calculated;Finally,the normal vector similarity is used as the constraint condition,and the improved region growth algorithm is used to extract the road surface.Two sections of urban road point cloud data are used for the experiment,and the results show that the accuracy CR,integrity CP and extraction quality Q of the road surface extraction results obtained by this method are greater than 94%,which verifies the effectiveness and robustness of this method.