一种基于极大曲率的机载LiDAR数据地形特征提取方法

A Method of Airborne LiDAR Data Terrain Feature Extraction Based on Maximum Curvature

对于庞大的点云数据来说,从中直接提取多种数据特征是相当困难的。考虑到基于数字化等高线数据和数字地面模型提取地形特征受内插误差影响,提出一种基于极大曲率的地形特征提取方法。对初始地面点云进行极大曲率估计;结合欧氏聚类方法进行地形特征点粗提取;对粗提取特征点进行粗糙度分析,得到精确特征点。顾及离散曲率的特征,即曲率越大越接近特征线,得到可靠的地形特征分割结果。该方法不需要人工干预,直接基于机载LiDAR点云数据进行处理。试验表明采用基于极大曲率分割的方法能从点云中自动提取比较完整并且准确的地形特征点。

It is rather difficult to extract various data feature directly from huge amount of point cloud.Considering the interpolation error influence from extracting terrain feature using the digital contour data and digital terrain model,a terrain feature extraction method based on the maximum curvature is presented in this paper.Firstly,maximum curvature for initial ground point cloud is estimated and the terrain feature points are crudely extracted combining Euclidean clustering method.Then,roughness analysis is proposed to simplify crude extraction point and obtain the accurate feature point.Considering the curvature features of discrete point cloud,the greater the point cloud curvature is that the closer the feature line is,this method has reliable coarse segmentation results for feature points.The experiment shows that maximum curvature segmentation method can extract integrated and precise terrain feature points from huge amount of point cloud automatically,it doesn’ t need manual intervention.