Advertisement

Stereo without search

  • Carlo Tomasi
  • Roberto Manduchi
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1064)

Abstract

In its traditional formulation, stereo correspondence involves both searching and selecting. Given a feature in one scanline, the corresponding scanline in the other image is searched for the positions of similar features. Often more than one candidate is found, and the correct one must be selected. The problem of selection is unavoidable because different features look similar to each other. Search, on the other hand, is not inherent in the correspondence problem. We propose a representation of scanlines, called intrinsic curves, that avoids search over different disparities. The idea is to represent scanlines by means of local descriptor vectors, without regard for where in the image a descriptor is computed, but without losing information about the contiguity of image points. In fact, intrinsic curves are the paths that the descriptor vector traverses as an image scanline is traversed from left to right. Because the path in the space of descriptors ignores image position, intrinsic curves are invariant with respect to disparity under ideal circumstances. Establishing stereo correspondences is then reduced to the selection of one among few match candidates, a task simplified by the contiguity information carried by intrinsic curves. We analyze intrinsic curves both theoretically and for real images in the presence of noise, brightness bias, contrast fluctuations, and moderate geometric distortion. We report preliminary experiments.

Keywords

Stereo Match Compatible Mapping Correspondence Problem Intrinsic Curve Match Candidate 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

References

  1. 1.
    R.D. Arnold and T.O. Binford. Geometric constraints in stereo vision. SPIE 238:281–292, 1978.Google Scholar
  2. 2.
    V.I. Arnold. Ordinary Differential Equations. MIT Press, 1990.Google Scholar
  3. 3.
    H.H. Baker and T.O. Binford. Depth from edge and intensity based stereo. IJCAI, 631–636, 1981.Google Scholar
  4. 4.
    P.N. Belhumeur and D. Mumford. A Bayesian treatment of the stereo correspondence problem using half-occluded regions. CVPR, 506–512, 1992.Google Scholar
  5. 5.
    A. Blake and C. Marinos. Shape from texture: Estimation, isotropy and moments. Artificial Intelligence, 45:323–380, 1990.Google Scholar
  6. 6.
    M. Campani and A. Verri. Motion analysis from first-order properties of optical flow. CVGIP: Image Understanding, 56(1):90–107, 1992.Google Scholar
  7. 7.
    D. Geiger, B. Ladendorf, and A. Yuille. Occlusions and binocular stereo. EECV, 425–433, 1992.Google Scholar
  8. 8.
    W.E.L. Grimson. Computational experiments with a feature based stereo algorithm. PAMI, 7(1):17–34, 1985.Google Scholar
  9. 9.
    W. Förstner. Image Matching. In R.M. Haralick and L.G. Shapiro, Computer and Robot Vision. Addison-Wesley, 1992.Google Scholar
  10. 10.
    S.S. Intille and A.F. Bobick. Disparity-space images and large occlusion stereo. ECCV, 179–186, 1994.Google Scholar
  11. 11.
    D.G. Jones and J. Malik. A computational framework for determining stereo correspondence from a set of linear spatial filters. EECV, 395–410, 1992.Google Scholar
  12. 12.
    D.G. Jones and J. Malik. Determining three-dimensional shape from orientation and spatial frequency disparities. EECV, 661–669, 1992.Google Scholar
  13. 13.
    K. Kanatani. Detection of surface orientation and motion from texture by a stereological technique. Artificial Intelligence, 23:213–237, 1984.Google Scholar
  14. 14.
    M.H. Kass. Computing stereo correspondence. Master's thesis, M.I.T., 1984.Google Scholar
  15. 15.
    J.J. Koenderink and A.J. Van Doorn. Geometry of binocular vision and a model for stereopsis. Biological Cybernetics, 21:29–35, 1976.Google Scholar
  16. 16.
    J.J. Little and W.E. Gillet. Direct evidence for occlusion in stereo and motion. ECCV, 336–340, 1990.Google Scholar
  17. 17.
    J. Malik and P. Perona. Preattentive texture discrimination with early vision mechanisms. JOSA — A, 7(5):923–932, 1990.Google Scholar
  18. 18.
    S.G. Mallat. A theory for multiresolution signal decomposition: The wavelet representation. PAMI, 11(7):674–693, 1989.Google Scholar
  19. 19.
    R. Manmatha. A framework for recovering affine transforms using points, lines or image brightness. CVPR, 141–146, 1994.Google Scholar
  20. 20.
    D. Marr and T. Poggio. Cooperative computation of stereo disparity. Science, 194:283–287, 1976.Google Scholar
  21. 21.
    G. Medioni and R. Nevatia. Segment-based stereo matching. CVGIP, 31:2–18, 1985.Google Scholar
  22. 22.
    Y. Ohta and T. Kanade. Stereo by intra-and inter-scanline search using dynamic programming. PAMI, 7(2):139–154, 1985.Google Scholar
  23. 23.
    S.B. Pollard, J.E. Mayhew, and G.P. Frisby. PMF: A stereo correspondence algorithm using a disparity gradient limit. Perception, 14:449–470, 1985.Google Scholar
  24. 24.
    J. Sato and R. Cipolla. Extracting the affine transformation from texture moments. ECCV, 165–172, 1994.Google Scholar
  25. 25.
    D.J. Struik. Lectures on Classical Differential Geometry. Dover, 1988.Google Scholar
  26. 26.
    B.J. Super and A.C. Bovik. Shape-from-texture by wavelet-based measurement of local spectral moments. CVPR, 296–301, 1992.Google Scholar
  27. 27.
    P.S. Toh and A.K. Forrest. Occlusion detection in early vision. ICCV, 126–132, 1990.Google Scholar
  28. 28.
    C. Tomasi and R. Manduchi. Stereo without search. Tech. Rep. STAN-CS-TR-95-1543, Stanford, 1995.Google Scholar
  29. 29.
    A. Verri and T. Poggio. Against quantitative optical flow. ICCV, 171–180, 1987.Google Scholar
  30. 30.
    J. Weng, N. Ahuja, and T.S. Huang. Matching two perspective views. PAMI, 14(8):806–825, 1992.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1996

Authors and Affiliations

  • Carlo Tomasi
    • 1
  • Roberto Manduchi
    • 1
  1. 1.Computer Science DepartmentStanford UniversityStanford

Personalised recommendations