Segmentation of 3D Point Cloud Data Based on Supervoxel Technique

R.S. Rampriya, R. Suganya
2020 Procedia Computer Science  
 Segmentation is a significant challenge posed to the computer vision applications dealing with the 3D point cloud data.In this paper, the bottomup 3D point cloud supervoxel technique is proposed for segmenting both outdoor and indoor scenes. Using this method, the dense point cloud data is r better computational efficiency. Initially, voxels and supervoxe data structure and voxelbased cloud connectivity algorith cut (CPC) of greedy cuts through a local connectivity is established
more » ... tween supervoxels realistic and meaningful objects (segments). The experiments have been performed on the part of the urban area around Vaihingen, Germany for the outdoor scene and RGB scene. It is demonstrated that our proposed method can achieve good results, especially for strict partitioning scenes. The quantitative comparison between various 3D point cloud segmentation methods is obtained and the proposed method confirms its efficiency and effectiveness.
doi:10.1016/j.procs.2020.04.045 fatcat:w5qdy4mkr5dm7m7ruzafim4xsi