自适应双边布料模拟滤波用于建筑物立面提取  

Adaptive bilateral cloth simulation filtering for building facade extraction

作　　者：潘凯能 PAN Kaineng(Geographic Information Engineering Brigade,Jiangxi Provincial Bureau of Geology,Nanchang,Jiangxi 330001,China)

机构地区：[1]江西省地质局地理信息工程大队,江西南昌330001

出　　处：《北京测绘》2025年第4期573-580,共8页Beijing Surveying and Mapping

摘　　要：从点云中提取精细的建筑物立面数据是计算建筑工程测量参数的前提,布料模拟滤波(CSF)是一种简单高效的滤波方法。针对CSF参数设置复杂且无法有效提取全部立面点的问题,本文提出自适应双边布料模拟滤波(AB-CSF),在计算立面点云法向量的前提下,设置两块相对运动的模拟布料;根据粗滤波结果,逐层构建四叉树,得到包含较大高程差的末层子节点;使用克拉斯卡-瓦立斯(Kruskal-Wallis)检验,评价末层子节点与理想节点的相似程度,并根据卡方检验值自动设置各节点参数。多个数据集的实验结果表明,本文算法有更小的Ⅰ类误差和总体误差,实现步骤简单可靠,鲁棒性高,在可接受的时间内显著提高了建筑物立面点云的提取精度。Extracting detailed building facade data from point clouds is a prerequisite for calculating building engineering measurement parameters.Cloth simulation filtering(CSF) is a simple and efficient filtering method.To address the issue of complex CSF parameter settings and the inability to effectively extract all facade points,this paper proposed an adaptive bilateral cloth simulation filtering(AB-CSF) method.Based on the calculation of the normal vectors of the facade point cloud,two mutually moving simulated cloths were set up.Using the results from coarse filtering,a quadtree was constructed layer by layer,obtaining the terminal child nodes with larger elevation differences.The Kruskal-Wallis test was employed to evaluate the similarity between the terminal child nodes and ideal nodes,and the parameters for each node were automatically set based on the Chi-square test value.Experimental results on multiple datasets show that the proposed algorithm achieves smaller Type I errors and overall errors,with a simple and reliable implementation,high robustness,and significantly improved building facade point cloud extraction accuracy within an acceptable time frame.

关 键 词：布料模拟滤波(CSF) 自适应参数 立面提取 建筑物点云 

分 类 号：P237[天文地球—摄影测量与遥感]