3D Reconstruction of Stereo Images for Interaction between Real and Virtual Worlds

  • Authors:
  • Hansung Kim;Seung-jun Yang;Kwanghoon Sohn

  • 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

Mixed reality is different from the virtual reality in thatusers can feel immersed in a space which is composed ofnot only virtual but also real objects. Thus, it is essentialto realize seamless integration and interaction of thevirtual and real worlds. We need depth information of thereal scene to synthesize the real and virtual objects. Wepropose a two-stage algorithm to find smooth and precisedisparity vector fields with sharp object boundaries in astereo image pair for depth estimation. Hierarchicalregion-dividing disparity estimation increases theefficiency and the reliability of the estimation process,and a shape-adaptive window provides high reliability ofthe fields around the object boundary region. At thesecond stage, the vector fields are regularized with aenergy model which produces smooth fields whilepreserving their discontinuities resulting from the objectboundaries. The vector fields are used to reconstruct 3Dsurface of the real scene. Simulation results show that theproposed algorithm provides accurate and spatiallycorrelated disparity vector fields in various kinds ofimages, and synthesized 3D models produce naturalspace where the virtual objects interact with the realworld as if they are in the same world.