利用RANSAC算法对建筑物立面进行点云分割

Segmentation of building facade point clouds using RANSAC

建筑物立面点云分割是车载激光扫描数据特征提取与建模的基础。本文将随机抽样一致性算法(Ran-dom Sampling Consensus)方法引入对点云的分割中,并在判断准则中引入了点云的r半径密度,消除了噪声的影响,同时建立角度和距离两个约束条件对平面分割结果进行优化,提取出了最终的建筑物立面特征平面。

Segmentation of building facade point clouds is the foundation of feature extraction and modeling from Vehicle-Borne LiDAR. In the paper, Random Sampling Consensus was introduced into the segmentation of LiDAR and r-radius point density was put forward to the estimation criterion, which aims to remove the discrete point outside the feature plane. Then two constraints of angle and distance were erected to unite the segmented planes which optimized the results.