Processing of a multiscale triangulated surface model of a 3D scene for a robotics application

Yue Liu; Denis Laurendeau

doi:10.1117/12.146642

Translator Disclaimer

23 June 1993 Processing of a multiscale triangulated surface model of a 3D scene for a robotics application

Yue Liu, Denis Laurendeau

Author Affiliations +

Proceedings Volume 2031, Geometric Methods in Computer Vision II; (1993) https://doi.org/10.1117/12.146642
Event: SPIE's 1993 International Symposium on Optics, Imaging, and Instrumentation, 1993, San Diego, CA, United States

Abstract

In this paper, we propose a multi-scale approach to 3-D scene modeling and processing. The basic idea of the proposed technique is to acquire a minimum amount of 3-D information using a random access range finder. The 3-D data is used to build a triangulation which is a coarse planar model of the scene and is called the coarse scale map (CSM). The triangulation is used as a guide to acquire more data in the areas of the scene that look the most `interesting' without changing the orientation of the camera, or to choose the position of the 3-D camera for the acquisition of the next best view. The additional data is added to the CSM triangulation and forms what is called the fine scale map (FSM). The FSM is also structured as a triangulation and is processed in order to segment the various components of a scene. The typical 3-D scene of interest is made of a wooden beam, an electrical insulator, and an electric cable. The goal of the 3-D image processing is to use the triangulation to find the beam and the insulator in the scene.

Citation Download Citation

Yue Liu and Denis Laurendeau "Processing of a multiscale triangulated surface model of a 3D scene for a robotics application", Proc. SPIE 2031, Geometric Methods in Computer Vision II, (23 June 1993); https://doi.org/10.1117/12.146642

ACCESS THE FULL ARTICLE