View Synthesis by Trinocular Edge Matching and Transfer

  • Authors:
  • Stephen Pollard;Maurizio Pilu;Sean Hayes;Adele Lorusso

  • Affiliations:
  • -;-;-;-

  • Venue:
  • WACV '98 Proceedings of the 4th IEEE Workshop on Applications of Computer Vision (WACV'98)
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents a novel automatic method for viewsynthesis (or image transfer) from a triplet ofuncalibrated images based on trinocular edge matchingfollowed by transfer by interpolation, occlusiondetection and correction and finally rendering. Theedge-based technique proposed here is of generalpractical relevance because it overcomes most of theproblems encountered in other approaches that eitherrely upon dense correspondence, work in projectivespace or need explicit camera calibration.Applications range from immersive media andteleconferencing, image interpolation for fast renderingand compression.