Dealing with segmentation errors in region-based stereo vision

• Ángeles López ^[1] ; Filiberto Pla ^[1]
  1. [1] Universitat Jaume I

     Universitat Jaume I

     Castellón, España

     
• Localización: CAEPIA'97: actas / coord. por Asociación Española de Inteligencia Artificial, Vicente J. Botti Navarro , 1997, ISBN 84-8498-765-5, págs. 317-326
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • In this paper an algorithm to deal with errors in region-based stereo vision is proposed. Graph-based techniques solve ambiguities when one region in an image corresponds to one region in the other image, but segmentation errors produce splitted regions that should be matched to the same region at the same time. A preprocessing stage is proposed, prior to the matching process, in order to create merged regions that could be matched to the existing regions. A main drawback of this method is solved here, where an appropiate similarity criterion to create the initial nodes in the association graph is proposed, which is different from the similarity criterion used in the rest of the matching process.