Accurate reconstruction of engineered models with surfaces of revolution

  • Authors:
  • Nithi Gupta;Subodh Kumar

  • Affiliations:
  • Indian Institute of Technology, Delhi, New Delhi;Indian Institute of Technology, Delhi, New Delhi

  • Venue:
  • Proceedings of the Eighth Indian Conference on Computer Vision, Graphics and Image Processing
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Scanners and other acquisition devices are commonly used to capture shapes of real objects as point clouds. Piecewise linear approximations of point clouds are then obtained by reconstruction algorithms. Generic reconstruction algorithms rely on assumptions like smoothness, uniform or dense sampling, etc. In general, they are unable to precisely regenerate salient features of the underlying model like edges and corners. We restrict our focus to engineered models that are formed by components having simple analytically defined shapes like planes, surfaces of revolution and extrusion. We employ a parameterization that allows linearization of constraints to recover these shapes. The regularity and relationships of the components are exploited to reconstruct the accurate shape with sharp edges from highly noisy point clouds.