一种基于机载激光雷达点云的建筑物识别方法

Building recognition method based on airborne LiDAR point cloud

以三维激光雷达数据为基础进行目标自动检测识别具有重要现实意义。针对传统的以人工判读识别目标的方法已经远不能满足从海量数据中确定目标的要求,文中提出一种基于机载激光雷达点云的建筑物识别方法。通过使用改进后的金字塔方法构建三维激光雷达点云,鉴于三维激光点云具有地物本身的语义信息,可根据点云的多尺度复杂地物特征实现三维激光点云数据的自动分类,对分类后的非地面点进行区域生长分割,通过点云与建筑物平面特征的匹配,确定建筑物的识别结果。整个方案实现过程清晰、简单,而且该方法的实现对计算机硬件配置要求不高,能够从海量三维激光雷达数据中快速、准确地识别类建筑物,正确率高达81.25%,在目标识别领域具有较高的应用价值。

The automatic target detection and recognition based on 3D LiDAR data is of great practical significance.The traditional manual interpretation and recognition methods have limitations to identify the target from massive data.To address this issue,a building recognition method based on airborne LiDAR point cloud is proposed.The improved pyramid method is used to construct 3D LiDAR point cloud.Because the 3D LiDAR point cloud has the semantic information of the ground object itself,its data can be classified automatically according to the multi-scale complex ground object features of the point cloud.The non-ground points can be segmented by region growing.The recognition result of the building is determined by matching the point cloud and the planar features of building.The implementation process is clear and easy.Furthermore,this method does not require high computer hardware configurations,and can quickly and accurately identify building-like structures in point cloud with an impressive accuracy rate of 81.25%.It has high application value in the field of target recognition.