Fully Automated and Stable Registration for Augmented Reality Applications

  • Authors:
  • Vincent Lepetit;Luca Vacchetti;Daniel Thalmann;Pascal Fua

  • Affiliations:
  • -;-;-;-

  • Venue:
  • ISMAR '03 Proceedings of the 2nd IEEE/ACM International Symposium on Mixed and Augmented Reality
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present a fully automated approach to camera registrationfor Augmented Reality systems. It relies on purelypassive vision techniques to solve the initialization and real-timetracking problems, given a rough CAD model of partsof the real scene. It does not require a controlled environment,for example placing markers. It handles arbitrarilycomplex models, occlusions, large camera displacementsand drastic aspect changes.This is made possible by two major contributions: Thefirst one is a fast recognition method that detects the knownpart of the scene, registers the camera with respect to it, andinitializes a real-time tracker, which is the second contribution.Our tracker eliminates drift and jitter by merging theinformation from preceding frames in a traditional recursivetracking fashion with that of a very limited number ofkey-frames created off-line. In the rare instances where itfails, for example because of large occlusion, it detects thefailure and reinvokes the initialization procedure.We present experimental results on several differentkinds of objects and scenes.