摘要
随着LIDAR技术在输电线路勘察设计的广泛应用,如何自动提取输电线路的房屋等建筑物成了一个新课题。以正确分类LIDAR点云为目标,综合利用不同类别目标点云的回波特征以及地形信息,本文提出了一种基于区域多次回波密度分析的LIDAR点云分类方法。该方法通过线性预测分离地面点和非地面点,同时将点云构建TIN获取封闭的等高线,利用等高线间的拓扑关系及形状相似性获得等高线族区域,统计每一区域的多次回波点云密度信息,通过建筑物和树木区域多次回波点云在区域密度上的巨大差异来识别出建筑物点云和树木点云。该方法充分利用了建筑物表面与植被间多次回波特性的差异,同时又不否定建筑物边缘同样存在的多次回波现象,通过等高线区域自适应的确定待识别目标的区域大小,弥补了传统LIDAR点云分类方法的不足,因而能够较其它方法更准确的分类LIDAR点云,为建筑物自动提取、树木自动提取等研究提供了保障,并通过试验验证了该方法的有效性。该研究为利用LIDAR技术进行输电线路GIS自动采集房屋提供了一种新方法,大大减少了外业采集房屋面积的工作量。
With a wide range of application for LIDAR technology in the transmission line survey and design, how to automatically extract the buildings has become a new topic. This paper presents a multi-echo point density based classification method for LIDAR point cloud by using the echo properties of different kind of objects and terrain information in order to classify the point cloud accurately. The proposed method is that first the terrain points and nonterrain points are distinguished by linear prediction, then the contour line are obtained from TIN constructed by the terrain points, finally the multi-echo point density of every contour line region which is determined by the topological relation of contour line and shape similarity is calculated, so the building points and tree points are recognized by the great difference of the regional multi-echo point densities of these two kind objects. This method makes up the traditional classification method for LIDAR points by utilizing the multi-echo point characteristic difference of the building surface and vegetation surface without ignoring the multi-echo phenomenon for building fringes. The results show that the new classification method can reach higher classification accuracy for automatic building extraction and automatic extraction of vegetation. The research provides a new approach for the use of LIDAR technology for automatic buildings extraction of Transmission Line GIS and greatly reduces the workload of site buildings collection.
出处
《电力勘测设计》
2017年第S1期161-165,共5页
Electric Power Survey & Design
关键词
LIDAR
点云分类
等高线族
区域回波密度
LIDAR
point cloud classification
contour cluster
regional echo density