Trinocular Stereo: A Real-Time Algorithm and Its Evaluation

  • Authors:
  • J. Mulligan;V. Isler;K. Daniilidis

  • Affiliations:
  • -;-;-

  • Venue:
  • SMBV '01 Proceedings of the IEEE Workshop on Stereo and Multi-Baseline Vision (SMBV'01)
  • Year:
  • 2001

Quantified Score

Hi-index 0.00

Visualization

Abstract

In telepresence applications each user is immersed in arendered 3D-world composed from representations transmittedfrom remote sites. The challenge is to compute denserange data at high frame rates, since participants cannoteasily communicate if the processing cycle or network latenciesare long. Moreover, errors in new stereoscopicviews of the remote 3D-world should be hardly perceptible.To achieve the required speed and accuracy, we use trinocularstereo, a matching algorithm based on the sum of modifiednormalized cross-correlations, and subpixel disparityinterpolation. To increase speed we use Intel IPL functionsin the pre-processing steps of background subtractionand image rectification as well as a four-processor parallelization.To evaluate our system we have developed a test-bedwhich provides a set of registered dense "ground-truth"laser data and image data from multiple views.