International Journal of Computer Vision

, Volume 51, Issue 1, pp 63–83 | Cite as

Self-Consistency and MDL: A Paradigm for Evaluating Point-Correspondence Algorithms, and Its Application to Detecting Changes in Surface Elevation

  • Yvan G. Leclerc
  • Q.-Tuan Luong
  • P. Fua


The self-consistency methodology is a new paradigm for evaluating certain vision algorithms without relying extensively on ground truth. We demonstrate its effectiveness in the case of point-correspondence algorithms and use our approach to predict their accuracy.

For point-correspondence algorithms, our methodology consists in applying independently the algorithm to subsets of images obtained by varying the camera geometry while keeping 3-D object geometry constant. Matches that should correspond to the same surface element in 3-D are collected to create statistics that are then used as a measure of the accuracy and reliability of the algorithm. These statistics can then be used to predict the accuracy and reliability of the algorithm applied to new images of new scenes.

An effective representation for these statistics is a scatter diagram along two dimensions: A normalized distance and a matching score. The normalized distance make the statistics invariant to camera geometry, while the matching score allows us to predict the accuracy of individual matches. We introduce a new matching score based on Minimum Description Length (MDL) theory, which is shown to be a better predictor of the quality of a match than the traditional Sum of Squared Distance (SSD) score.

We demonstrate the potential of our methodology in two different application areas. First, we compare different point-correspondence algorithms, matching scores, and window sizes. Second, we detect changes in terrain elevation between 3-D terrain models reconstructed from two sets of images taken at a different time.

We finish by discussing the application of self-consistency to other vision problems.

consistency stereo minimal description length correspondences 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. Anandan, P. 1989. A computational framework and an algorithm for the measurement of visual motion. International Journal of Computer Vision, 2:283-310.Google Scholar
  2. Ayache, N. 1991. Artificial Vision for Mobile Robots. MIT Press: Cambridge, MA.Google Scholar
  3. Ayache, N. and Lustman, F. 1987. Fast and reliable passive trinocular stereovision. In International Conference on Computer Vision, pp. 422-427.Google Scholar
  4. Bejanin, M., Huertas, A., Medioni, G., and Nevatia, R. 1994. Model validation for change detection. In WACV94, pp. 160-167.Google Scholar
  5. Boult, T., Micheals, R., Erkan, A., Lewis, P., Powers, C., Qian, C., and Yin, W. 1998. Frame-rate multi-body tracking for surveillance. In DARPA98, pp. 305-313.Google Scholar
  6. Cho, K., Meer, P., and Cabrera, J. 1997. Performance assessment through bootstrap. IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(11):1185-1198.Google Scholar
  7. Csurka, G., Zeller, C., Zhang, Z., and Faugeras, O. 1996. Characterizing the uncertainty of the fundamental matrix. In CVGIP-IU.Google Scholar
  8. Faugeras, O., Fua, P., Hotz, B., Ma, R., Robert, L., Thonnat, M., and Zhang, Z. 1992. Quantitative and qualitative comparison of some area and feature-based stereo algorithms. In International Workshop on Robust Computer Vision: Quality of Vision Algorithms, W. Forstner and S. Ruwiedel (Eds.). Karlsruhe, Germany, pp. 1-26.Google Scholar
  9. Forstner, W. 1982. On the geometric precision of digital correlation. In International Archives of Photogrammetry and Remote Sensing, Vol. 24-III, Helsinki, pp. 176-189.Google Scholar
  10. Førstner, W. 1994. Diagnostics and performance evaluation in computer vision. In Performance Versus Methodology in Computer Vision, NSF/ARPA Workshop, Seattle, WA.Google Scholar
  11. Fua, P. 1991. Combining stereo and monocular information to compute dense depth maps that preserve depth discontinuities. In Int. Joint Conf. on AI, Sydney, Australia, pp. 1292-1298.Google Scholar
  12. Fua, P. and Leclerc, Y.G. 1995. Object-centered surface reconstruction: Combining multi-image stereo and shading. International Journal of Computer Vision, 16:35-56.Google Scholar
  13. Horn, B.K.P. and Brooks, M.J. 1989. Shape fromShading. MITPress: Cambridge, MA.Google Scholar
  14. Huertas, A. and Nevatia, R. 2000. Detecting changes in aerial views of man-made structures. IVC, 18(8):583-596.Google Scholar
  15. Kanade, T. and Okutomi, M. 1994. A stereo matching algorithm with an adaptive window: Theory and experiment. IEEE Transactions on Pattern Analysis and Machine Intelligence, 16(9):920- 932.Google Scholar
  16. Leclerc, Y.G. 1989. Constructing simple stable descriptions for image partitioning. International Journal of Computer Vision, 3(1):73-102.Google Scholar
  17. Leclerc, Y.G. and Fischler, M.A. 1992. An optimization-based approach to the interpretation of single line drawings as 3d wire frames. International Journal of Computer Vision, 9(2):113- 136.Google Scholar
  18. Leclerc, Y., Luong, Q.-T., and Fua, P. 1999. Self-consistency: A novel approach to characterizing the accuracy and reliability of point correspondence algorithms. In Proceedings of the One-day Workshop on Performance Characterisation and Benchmarking of Vision Systems, Las Palmas de Gran Canaria, Canary Islands, Spain.Google Scholar
  19. Lillestrand, R.L. 1972. Techniques for change detection. TC, 21(7):654-659.Google Scholar
  20. Matthies, L. 1992. Stereo vision for planetary rovers: Stochastic modeling to near real-time implementation. International Journal of Computer Vision, 8(1):71-91.Google Scholar
  21. Mohr, R., Veillon, F., and Quan, L. 1993. Relative 3d reconstruction using multiple uncalibrated images. In Conference on Computer Vision and Pattern Recognition, NYC, pp. 543-548.Google Scholar
  22. Quam, L.H. 1971. Computer comparison of pictures. PhD Thesis, Stanford University.Google Scholar
  23. Rissanen, J. 1989. Minimum-description length principle. Encyclopedia of Statistical Sciences, 3(1):73-102.Google Scholar
  24. Rosin, P.L. 1998. Thresholding for change detection. In ICCV98, pp. 274-279.Google Scholar
  25. Sarkar, S. and Boyer, K.L. 1998. Quantitative measures of change based on feature organization: Eigenvalues and eigenvectors. Computer Vision and Image Understanding, 71(1):110-136.Google Scholar
  26. Sugihara, K. 1986. Machine interpretation of line drawings. MIT Press.Google Scholar
  27. Szeliski, R. 1999. Prediction error as a quality metric for motion and stereo. In ICCV99, Corfu, Greece.Google Scholar
  28. Szeliski, R. and Kang, S.B. 1994. Recovering 3d shape and motion from image streams using nonlinear least squares. JVCIR, pp. 10- 28.Google Scholar
  29. Szeliski, R. and Zabih, R. 1999. An experimental comparison of stereo algorithms. In Proceedings of theVision Algorithms: Theory and Practice Workshop (ICCV99), Corfu, Greece.Google Scholar
  30. Torr, P.H.S. and Zissermann, A. 1997. Performance characterization of fundamental matrix estimation under image degradation. Machine Vision and Applications, 9:321-333.Google Scholar
  31. Yachida, M., Kitamura, Y., and Kimachi, M. 1986. Trinocular vision: New approach for correspondence problem. In International Conference on Pattern Recognition, pp. 1041-1044.Google Scholar
  32. Yi, S., Haralick, R.M., and Shapiro, L.G. 1994. Error propagation in machine vision. Machine Vision and Applications, 7:93- 114.Google Scholar

Copyright information

© Kluwer Academic Publishers 2003

Authors and Affiliations

  • Yvan G. Leclerc
    • 1
  • Q.-Tuan Luong
    • 1
  • P. Fua
    • 2
  1. 1.Artificial Intelligence CenterSRI InternationalMenlo ParkUSA
  2. 2.EPFL, cLausanneSwitzerland

Personalised recommendations