一种建筑物点云轮廓线的自动提取方法  

An Automatic Building Boundary Extraction Method of TLS Data

作　　者：秦家鑫[1] 万幼川[1] 何培培[1] 陈茂霖[1] 卢维欣[1] 王思颖[1] 

机构地区：[1]武汉大学遥感信息工程学院,武汉430079

出　　处：《遥感信息》2015年第4期3-7 13,13,共6页Remote Sensing Information

基　　金：863计划(2013AA122104-3);博士点基金(20130141130003);国家科技支撑(2012BAJ15B04)

摘　　要：针对地面激光点云包含大量冗余数据、特征信息不明显等缺点,提出了一种自动提取建筑物点云轮廓线的方法。首先基于主成分分析和熵函数计算每个点的最佳邻域,再根据几何位置关系,滤除散乱点和平面中的点,保留轮廓线点云。针对不同地面激光点云数据,无需反复调整阈值。实验证明,该方法提取的建筑物点云轮廓线清晰完整,与现有方法提取的结果相比正确率有一定的提高,且冗余点云几乎全部被滤除,自动化程度高,具有良好的适用性。In view of the redundancy of point cloud data and the shortcoming of unconspicuous features,this paper presents an automatic method of building boundary extraction.Firstly,the optimal neighborhood is acquired according to the principal component analysis and entropy function.Then,scattered points are deleted and boundaries of planes are extracted by geometry information.Experimental results show that this method can not only extract boundary completely and filter almost all improper point cloud data,but also has a high applicability and automaticity without adjusting threshold to different TLS data.

关 键 词：地面激光扫描仪 主成分分析 熵函数 最佳邻域 轮廓线提取 LIDAR 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]