一种随机抽样一致性算法的建筑物平面点云提取方法  

RANSAC LiDAR Building Extraction Method with Constraints

作　　者：何明 李勇[1] 方秀琴[1] 周嘉俊 HE Ming;LI Yong;FANG Xiuqin;ZHOU Jiajun(School of Earth Sciences and Engineering,Hohai University,Nanjing 211100,China)

机构地区：[1]河海大学地球科学与工程学院,南京211100

出　　处：《遥感信息》2018年第3期104-107,共4页Remote Sensing Information

基　　金："十三五"国家重点研发计划项目(2016YFA0601504);国家自然科学基金(41101374);江苏省自然科学基金(BK20161504);测绘遥感信息工程国家重点实验室开放基金(14E02);中央高校基本科研业务费专项资金资助(2015B16614)

摘　　要：针对传统RANSAC算法在提取建筑物点云时存在迭代次数过多、精确度不高等缺点,提出了一种改进的带约束条件的RANSAC算法,结合三维形态学腐蚀运算对建筑物点云进行自动提取。首先利用带约束条件的RANSAC算法提取初始建筑物点云平面,再利用腐蚀运算去除误分类的部分低矮植被点云。实验结果表明,该方法在得到较高的精度和鲁棒性的情况下,还大幅度减少了运算量,提高了建筑物点云提取的速度。In common RANSAC（random sampling consensus）algorithm for building extraction from point clouds,the number of iterations is too large and the accuracy is not very high.To overcome these problems,this paper presents an improved constrained RANSAC algorithm combined with 3 D morphological erosion operation to extract building automatically.Firstly,the RANSAC algorithm with constraints is used to extract the initial building point cloud plane.And then,the erosion operation is used to remove some low vegetation point clouds which are not classified correctly.Finally,experiment demonstrates the accuracy and robustness of the method.This method greatly reduces the amount of computation and accelerates the speed of building extraction.

关 键 词：RANSAC 建筑物点云 约束条件 提取 腐蚀运算 

分 类 号：P237[天文地球—摄影测量与遥感]