采用车载激光雷达的带状地物矢量化与结构特征提取

Vectorization and Structural Feature Extraction of Strip Objects Using Vehicle-Mounted LiDAR

针对车载激光点云带状地物形状多样、难以用规则语义信息矢量化,以及结构语义信息提取研究较少的问题,提出一种基于Ribbon Snake模型的车载激光雷达带状地物(道路边界、实线型标线、铁轨)矢量化与结构特征提取方法.首先,通过格网剖分构建点云特征图,利用Ribbon Snake模型提取带状地物矢量化数据;然后,分析不同道路和铁路场景的结构特征,生成具有准确几何和拓扑结构信息的三维矢量数据和属性数据.实验表明:该方法能够准确地提取带状地物矢量化与结构信息,实现不同场景下带状地物的有效完整描述.

Due to the diverse shapes of strip objects,it is difficult to use regular semantic information for vectorization,and there is less research on structural semantic information extraction,a novel method for vectorization and structural feature extraction of strip objects(road boundary,solid line marking,railroad track)from vehicle-mounted LiDAR based on Ribbon Snake model is proposed.Firstly,a point cloud feature map was generated based on the salient strip points,and the Ribbon Snake model is used to extract the vectorization boundaries of the strip objects.Then analyze the structural characteristics of different road and railway scenes to generate 3D vector data and attribute data with accurate geometric and topological structure information.Experiments showed that the proposed method has accurately gained vectorization and structure information of strip objects,and achieved effective and complete description of strip objects in various scenes.