一种从机载LiDAR点云中分割建筑物顶面的方法

A Method for Building Roof Segmentation from Airborne LiDAR Point Cloud

建筑物的三维重建是根据现有数据自动恢复高精度建筑物三维模型的过程,它是摄影测量,计算机视觉以及模式识别等领域的热点问题。针对这一问题,提出一种从机载LiDAR(light detection and ranging)点云中分割建筑物顶面的方法。首先通过欧式聚类来得到每一栋建筑物的点云,随后采用L0梯度最小化以及图割优化相结合的全局最优算法来获得平面分割结果。并在德国Vaihingen以及中国杭州数据集上进行试验,证明该方法可以从LiDAR点云中自动获得高精度的平面分割结果,且具备强鲁棒性。

3D Building reconstruction is to automatically reconstruct the building model with high precision from the existing data.It is an important topic in the fields of photogram‐metry,computer vision and pattern recognition.To tackle this problem,this paper propose a method for segmenting the top surface of buildings from airborne LiDAR(light detection and ranging)point clouds,obtaining point clouds for each building through Euclidean clustering,A two-stage method comprised of L0 gradient minimization and graph-cut optimization is applied to obtain the roof plane segmentation results.We made experiments on the datasets of Vaihingen in Germany and Hangzhou in China.Experimental results show that our method can automatically obtain roof plane segmentation results with high quality and strong robustness from airborne LiDAR point clouds.