Learning Class Specific Graph Prototypes

  • Shengping Xia
  • Edwin R. Hancock
Part of the Lecture Notes in Computer Science book series (LNCS, volume 5716)


This paper describes how to construct a graph prototype model from a large corpus of multi-view images using local invariant features. We commence by representing each image with a graph, which is constructed from a group of selected SIFT features. We then propose a new pairwise clustering method based on a graph matching similarity measure. The positive example graphs of a specific class accompanied with a set of negative example graphs are clustered into one or more clusters, which minimize an entropy function. Each cluster is simplified into a tree structure composed of a series of irreducible graphs, and for each of which a node co-occurrence probability matrix is obtained. Finally, a recognition oriented class specific graph prototype (CSGP) is automatically generated from the given graph set. Experiments are performed on over 50K training images spanning ~500 objects and over 20K test images of 68 objects. This demonstrates the scalability and recognition performance of our model.


Learn Class Sift Feature Model View View Cluster Large Image Database 
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.
    Bonev, B., Escolano, F., Lozano, M.A., Suau, P., Cazorla, M.A., Aguilar, W.: Constellations and the unsupervised learning of graphs. GbRPR 14(1), 340–350 (2007)zbMATHGoogle Scholar
  2. 2.
    Chung, F.: Spectral graph theory. American Mathematical Society, Providence (1997)zbMATHGoogle Scholar
  3. 3.
    Crandall, D.J., Huttenlocher, D.P.: Weakly supervised learning of part-based spatial models for visual object recognition. In: Leonardis, A., Bischof, H., Pinz, A. (eds.) ECCV 2006. LNCS, vol. 3951, pp. 16–29. Springer, Heidelberg (2006)CrossRefGoogle Scholar
  4. 4.
    Everingham, M., Gool, L.V., Williams, C., Winn, J., Zisserman, A.: Overview and results of classification challenge, 2007. In: The PASCAL VOC 2007 Challenge Workshop, in conj. with ICCV (2007)Google Scholar
  5. 5.
    Fawcett, T.: Roc graphs: Notes and practical considerations for researchers (2004),
  6. 6.
    Ferrari, V., Tuytelaara, T., Van-Gool, L.: Simultaneous object recognition and segmentation from single or multiple model views. IJCV 67(2), 159–188 (2006)CrossRefGoogle Scholar
  7. 7.
    Jiang, X., Munger, A., Bunke, H.: On median graphs: properties, algorithms, and applications. PAMI 23(10), 1144–1151 (2001)CrossRefGoogle Scholar
  8. 8.
    Li, F.F., Perona, P.: A bayesian hierarchical model for learning natural scene categories. CVPR 2(2), 524–531 (2005)Google Scholar
  9. 9.
    Lowe, D.: Local feature view clustering for 3d object recognition. CVPR 2(1), 1682–1688 (2001)Google Scholar
  10. 10.
    Lowe, D.: Distinctive image features from scale-invariant key points. IJCV 60(2), 91–110 (2004)CrossRefGoogle Scholar
  11. 11.
    Nister, D., Stewenius, H.: Scalable recognition with a vocabulary tree. CVPR 2(2), 2161–2168 (2006)Google Scholar
  12. 12.
    Rothganger, F., Lazebnik, S., Schmid, C., Ponce, J.: 3d object modeling and recognition using local affine-invariant image descriptors and multi-view spatial constraints. IJCV 66(3), 231–259 (2006)CrossRefGoogle Scholar
  13. 13.
    Schonemann, P.: A generalized solution of the orthogonal procrustes problem. Psychometrika 31(3), 1–10 (1966)MathSciNetCrossRefzbMATHGoogle Scholar
  14. 14.
    Sivic, J., Russell, B.C., Efros, A.A., Zisserman, A., Freeman, W.T.: Discovering objects and their location in images. ICCV 1(1), 872–877 (2005)Google Scholar
  15. 15.
    Todorovic, S., Ahuja, N.: Unsupervised category modeling, recognition and segmentation in images. PAMI (in press)Google Scholar
  16. 16.
    Torralba, A., Fergus, R., Weiss, Y.: Small codes and large image databases for recognition. In: CVPR (2008)Google Scholar
  17. 17.
    Torsello, A., Hancock, E.: Learning shape-classes using a mixture of tree-unions. PAMI 28(6), 954–967 (2006)CrossRefGoogle Scholar
  18. 18.
    Xia, S., Hancock, E.R.: 3d object recognition using hyper-graphs and ranked local invariant features. In: da Vitoria Lobo, N., Kasparis, T., Roli, F., Kwok, J.T., Georgiopoulos, M., Anagnostopoulos, G.C., Loog, M. (eds.) S+SSPR 2008. LNCS, vol. 5342, pp. 117–126. Springer, Heidelberg (2008)CrossRefGoogle Scholar
  19. 19.
    Xia, S., Hancock, E.R.: Clustering using class specific hyper graphs. In: da Vitoria Lobo, N., Kasparis, T., Roli, F., Kwok, J.T., Georgiopoulos, M., Anagnostopoulos, G.C., Loog, M. (eds.) S+SSPR 2008. LNCS, vol. 5342, pp. 318–328. Springer, Heidelberg (2008)CrossRefGoogle Scholar
  20. 20.
    Xia, S.P., Liu, J.J., Yuan, Z.T., Yu, H., Zhang, L.F., Yu, W.X.: Cluster-computer based incremental and distributed rsom data-clustering. ACTA Electronica sinica 35(3), 385–391 (2007)Google Scholar
  21. 21.
    Xia, S.P., Ren, P., Hancock, E.R.: Ranking the local invariant features for the robust visual saliencies. In: ICPR 2008 (2008)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Shengping Xia
    • 1
  • Edwin R. Hancock
    • 2
  1. 1.ATR Lab, School of Electronic Science and EngineeringNational University of Defense TechnologyChangshaP.R. China
  2. 2.Department of Computer ScienceUniversity of YorkYorkUK

Personalised recommendations