Pose Estimation of Free-Form 3D Objects without Point Matching using algebraic surface Models

Jean-Philippe Tarel, Hakan Civi and David Cooper

Abstract

This paper presents new approaches to two fundamental 3D vision problems: 3D surface representation, and fast extraction of geometric information from this representation, particularly 3D alignment. For surface representation by implicit algebraic surfaces, a non-iterative, robust, repeatable and approximately least squares fitting algorithm, 3D 3L, is proposed. The alignment problem is solved by formulating an intrinsic coordinate system determined by tensor contractions of the surface representation parameters.

Reference

@inproceedings{jpt-mb3ia,
   author    = {Jean-Philippe Tarel and Hakan Civi and David B. Cooper},
   title     = {Pose Estimation of Free-Form 3D Objects without Point Matching using algebraic surface Models},
   booktitle = {Proceedings of IEEE Worshop Model Based 3D Image Analysis},  
   address   = {Mumbai, India},
   date      = {January 3},
   pages     = {13--21},
   year	     = {1998},
   note      = {http://perso.lcpc.fr/tarel.jean-philippe/mb3ia.html}
}

Compressed postscript file (469 Kb)
Pdf file (469 Kb)
(c) IEEE