Real-time surface of revolution reconstruction on dense SLAM

Liming Yang (1, 2) , Hideaki Uchiyama (3) , Jean-Marie Normand (1, 4, 2) , Guillaume Moreau (1, 4, 2) , Hajime Nahagara (3) , Rin-Ichiro Taniguchi (3)

1 ECN - École Centrale de Nantes
2 CRENAU - Centre de recherche nantais Architectures Urbanités
3 Kyushu University
4 Hybrid - 3D interaction with virtual environments using body and mind

We present a fast and accurate method for reconstructing surfaces of revolution (SoR) on 3D data and its application to structural modeling of a cluttered scene in real-time. To estimate a SoR axis, we derive an approximately linear cost function for fast convergence. Also, we design a framework for reconstructing SoR on dense SLAM. In the experiment results, we show our method is accurate, robust to noise and runs in real-time.

Mots clés: 3D reconstruction; surfaces-of-revolution; point clouds

Domaines: Synthèse d'image et réalité virtuelle [cs.GR]

Liming Yang, Hideaki Uchiyama, Jean-Marie Normand, Guillaume Moreau, Hajime Nahagara, et al.. Real-time surface of revolution reconstruction on dense SLAM. 3D Vision Conference 2016, Oct 2016, Palo Alto, United States. ⟨hal-01389060⟩