Detection of Space Connectivity from Point Cloud for Stair Reconstruction

EasyChair Preprint no. 1001

Abstract

Stairs are common features in indoor environments that play an important role in structured indoor reconstruction. Despite the rapid development of indoor reconstruction from point clouds, the problem of stair reconstruction is far from being resolved. The current staircase detection methods based on partial sensory data are not suitable for staircase modeling because they determine only the geometric structure of stair steps. As staircase model is among the most important components of an indoor model and stair space is a subspace of indoor space, the approach for indoor modeling should be able to determine the spatial extent of the stair connection space as well as its relationship with other subspaces. In this study, the semantic definition of stair space and stair connection space are defined and a novel stair reconstruction method by detecting space connectivity from the point cloud is proposed for stair reconstruction. The proposed method is verified on four datasets with different stair types. The results indicate that the proposed method is well suited for staircase modeling in multi-story indoor environments.

Keyphrases: indoor space, point cloud, Space Connectivity, Stair Connection Space, Stair Reconstruction

@Booklet{EasyChair:1001,
  author = {Fan Yang and Yifan Liang and Dalin Li and Fei Su and Haihong Zhu and Xinkai Zuo and Lin Li},
  title = {Detection of Space Connectivity from Point Cloud for Stair Reconstruction},
  howpublished = {EasyChair Preprint no. 1001},

  year = {EasyChair, 2019}}