一种车载激光点云数据中道路自动提取方法

发布时间：2018-07-31 11:56

【摘要】：针对车载移动测量系统数据采集特点,构建车载激光点云扫描线索引,提出了一种基于扫描线索引的道路路面与路边点云稳健分类法。首先通过分析扫描线上不同地物剖面的空间分布特征,进行剖面激光点生长聚类,形成完整的地物剖面目标点集;然后根据点集的几何特征因子判断点集类型;最后利用相邻多条扫描线上路边点分布规律进行去噪。对车载移动测量系统获取的两份点云数据进行实验,路面与路边提取的平均完整率分别为94.4%、86%,平均准确率分别为98.9%、99.1%。实验分析表明,该方法能有效减少粗糙路面点的错误分类,适应不同的道路路边条件,降低独立地物对路边提取的干扰。
[Abstract]:According to the data acquisition characteristics of the vehicle mobile measurement system, the laser scanning line index is constructed, and a robust classification method of road surface and roadside point cloud is proposed based on the scanning line index. Firstly, by analyzing the spatial distribution characteristics of different feature sections on the scanning line, the laser point growth clustering is carried out to form a complete target point set of the terrain profile, and then the point set type is judged according to the geometric characteristic factors of the point set. Finally, the noise reduction is carried out by using the distribution rule of roadside points on several adjacent scan lines. Two points cloud data obtained by vehicle moving measurement system are tested. The average integrity rate of road surface and roadside extraction is 94.4 / 86, and the average accuracy is 98.9 / 99.1, respectively. The experimental results show that this method can effectively reduce the error classification of rough road surface points, adapt to different road roadside conditions, and reduce the interference of independent ground objects to roadside extraction.
【作者单位】：山东科技大学测绘科学与工程学院;山东科技大学海洋工程研究院;
【基金】：国家重大仪器设备开发专项(2013YQ120343) 海岛(礁)测绘技术国家测绘地理信息局重点实验室项目山东科技大学人才引进科研启动基金项目资助(2016RCJJ004)~~
【分类号】：P235

【相似文献】