Statistically Valid Graph Representations of Scale-Space Geometry

  • Tomoya Sakai
  • Atsushi Imiya
Part of the Lecture Notes in Computer Science book series (LNCS, volume 5099)


This paper presents a statistical scale-selection criterion for graph representations derived from differential geometric features of a greyscale image in a Gaussian scale space. The image gradient in scale space derives hierarchical and topological relationships among the bright and dark components in the image. These relationships can be represented as a tree and a skeleton-like graph, respectively. Since the image at small scales contains invalid geometric features due to noise and numerical errors, a validation scheme is required for the detected features. The presented scale-selection criterion allows us to identify the valid features used for the graph representations with statistical confidence.


Singular Point Probability Density Function Saddle Point Noise Image Scale Space 
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.


  1. 1.
    Witkin, A.P.: Scale space filtering. In: Proc. of 8th IJCAI, pp. 1019–1022 (1986)Google Scholar
  2. 2.
    Koenderink, J.J.: The structure of images. Biological Cybernetics 50, 363–370 (1984)zbMATHCrossRefMathSciNetGoogle Scholar
  3. 3.
    Lindeberg, T.: Scale-Space Theory in Computer Vision. Kluwer, Boston (1994)Google Scholar
  4. 4.
    Weickert, J., Ishikawa, S., Imiya, A.: Linear Scale-Space has First been Proposed in Japan. Journal of Mathematical Imaging and Vision 10, 237–252 (1999)zbMATHCrossRefMathSciNetGoogle Scholar
  5. 5.
    Zhao, N.-Y., Iijima, T.: A theory of feature extraction by the tree of stable view-points. IEICE Japan, Trans. D. J68-D, 1125–1135 (1985) (in Japanese)Google Scholar
  6. 6.
    Lifshitz, L.M., Pizer, S.M.: A multiresolution hierarchical approach to image segmentation based on intensity extrema. IEEE Trans. Pattern Analysis and Machine Intelligence 12(6), 529–540 (1990)CrossRefGoogle Scholar
  7. 7.
    Griffin, L.D., Colchester, A.: Superficial and deep structure in linear diffusion scale space: Isophotes, critical points and separatrices. Image and Vision Computing 13(7), 543–557 (1995)CrossRefGoogle Scholar
  8. 8.
    Olsen, O.F., Nielsen, M.: Generic events for the gradient squared with application to multi-scale segmentation. In: ter Haar Romeny, B.M., Florack, L.M.J., Viergever, M.A. (eds.) Scale-Space 1997. LNCS, vol. 1252, pp. 101–112. Springer, Heidelberg (1997)Google Scholar
  9. 9.
    Florack, L.M.J., Kuijper, A.: The topological structure of scale-space images. Journal of Mathematical Imaging and Vision 12(1), 65–79 (2000)zbMATHCrossRefMathSciNetGoogle Scholar
  10. 10.
    Kuijper, A., Florack, L.M.J., Viergever, M.A.: Scale space hierarchy. Journal of Mathematical Imaging and Vision 18(2), 169–189 (2003)zbMATHCrossRefMathSciNetGoogle Scholar
  11. 11.
    Sakai, T., Imiya, A.: Gradient structure of image in scale space. Journal of Mathematical Imaging and Vision 28(3), 243–257 (2007)CrossRefMathSciNetGoogle Scholar
  12. 12.
    Roberts, S.J.: Parametric and non-parametric unsupervised cluster analysis. Pattern Recognition 30(2), 261–272 (1997)CrossRefGoogle Scholar
  13. 13.
    Nakamura, E., Kehtarnavaz, N.: Determining number of clusters and prototype locations via multi-scale clustering. Pattern Recognition Letters 19(14), 1265–1283 (1998)zbMATHCrossRefGoogle Scholar
  14. 14.
    Leung, Y., Zhang, J.-S., Xu, Z.-B.: Clustering by scale-space filtering. IEEE Trans. on Pattern Analysis and Machine Intelligence 22(12), 1396–1410 (2000)CrossRefGoogle Scholar
  15. 15.
    Sakai, T., Imiya, A., Komazaki, T., Hama, S.: Critical scale for unsupervised cluster discovery. In: Perner, P. (ed.) MLDM 2007. LNCS (LNAI), vol. 4571, pp. 218–232. Springer, Heidelberg (2007)CrossRefGoogle Scholar
  16. 16.
    Muybridge, E.: Animal Locomotion (1887)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2008

Authors and Affiliations

  • Tomoya Sakai
    • 1
  • Atsushi Imiya
    • 1
  1. 1.Institute of Media and Information TechnologyChiba UniversityJapan

Personalised recommendations