基于几何特征约束的建筑物点云配准算法  

Geometric Feature Constraint Based Algorithm for Building Scanning Point Cloud Registration

作　　者：郑德华[1] 岳东杰[1] 岳建平[1] 

机构地区：[1]河海大学土木工程学院,江苏南京210098

出　　处：《测绘学报》2008年第4期464-468,共5页Acta Geodaetica et Cartographica Sinica

基　　金：国家科技支撑计划项目(2006BAG04B04);江苏省交通科学研究计划项目(05Y01)

摘　　要：针对人工建筑物表面存在的几何特征关系提出了基于几何特征约束的建筑物点云配准算法,根据点云数据中平面与平面重合关系,推导点在平面上和平面法线平行的2种线性不等约束条件。在6独立参数模型中增加几何特征约束的不等约束条件组成了附有约束条件的配准模型。通过对建筑物3维激光扫描点云数据的采集和处理,详细分析了几何特征约束配准算法的处理结果。试验结果分析表明几何特征约束条件可以合理地改善3维空间转换参数解算结果,提出的配准模型较适合于人工建筑物点云数据的配准。Aiming at tbe geometric characteristic in building surface, geometric feature constraint based algorithm for building point cloud registration is provided in the paper. According to superposition between two planes, by linear ization the two linear inequations are derived from the constraint condition of point on-plane and plane-normal-direc tion parallel. The registration mode with geometric feature constraint is composed of 6 parameter registration model and constraint condition linear inequation, which is calculated by iterative algorithm. Through data processing of building point cloud, the results of registration with geometric feature constraint are analyzed. The experimental result indicates that constraint condition be capable of improving the effect of registration, and the registration algorithm be suitable for the registration of artificial building point cloud.

关 键 词：几何特征约束 建筑物扫描 点云数据 配准算法 

分 类 号：P258[天文地球—测绘科学与技术]