Building outline extraction from ALS point clouds using medial axis transform descriptors

作者：

Highlights：

• Automatic building outline extraction, required by various spatial modeling applications and mapping, remains challenging due to object and data complexities.

• Skeletonization encodes topological information of point cloud data that can be used to extract object boundaries but it has traditional drawback due to its instability to small perturbations.

• We proposed a robust medial axis transform (MAT) based skeletonization method and designed a novel segmentation procedure for MAT points that allows to efficiently identify building corner points.

摘要

•Automatic building outline extraction, required by various spatial modeling applications and mapping, remains challenging due to object and data complexities.•Skeletonization encodes topological information of point cloud data that can be used to extract object boundaries but it has traditional drawback due to its instability to small perturbations.•We proposed a robust medial axis transform (MAT) based skeletonization method and designed a novel segmentation procedure for MAT points that allows to efficiently identify building corner points.

论文关键词：Skeleton,Point cloud,Building outline,Medial axis transform,Segmentation

论文评审过程：Received 7 November 2019, Revised 22 April 2020, Accepted 13 May 2020, Available online 25 May 2020, Version of Record 30 May 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107447