Advertisement

Constructing a Novel Pos-neg Manifold for Global-Based Image Classification

  • Rong Zhu
  • Jianhua Yang
  • Yonggang Li
  • Jie Xu
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8346)

Abstract

For the task of global-based image classification, we construct an image manifold, i.e., a pos-neg manifold, based on the solving strategies of two-class classification problem, which includes a positive sub-manifold and a negative one. We also present an improved globular neighborhood based locally linear embedding (an improved GNLLE) algorithm, fully taking account of the big differences between the positive and negative category images, thus the data distance calculation defined in the high-dimensional space can be translated into the one on the image manifold with lower dimensionality. Moreover, to simplify the distance measure between two nonlinear sub-manifolds, we put forward a clustering-based method to determine a manifold center for each sub-manifold. Experimental results on the real-world Web images show that the proposed method can improve the classification performance significantly.

Keywords

Global-based image classification image manifold two-class classification problem distance measure manifold center 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Kamde, P.M., Algur, S.P.: A Survey on Web Multimedia Mining. Int. J. Multimedia & Its Applications 3(3), 72–84 (2011)CrossRefGoogle Scholar
  2. 2.
    Bhatt, C.A., Kankanhulli, M.S.: Multimedia Data Mining: State of the Art and Challenges. Multimedia Tools and Applications 51(1), 35–76 (2011)CrossRefGoogle Scholar
  3. 3.
    Zeng, Z.Y., Yao, Z.Q., Liu, S.G.: An Efficient and Effective Image Representation for Region-Based Image Retrieval. In: 2nd International Conference on Interaction Sciences: Information Technology, Culture and Human, pp. 429–434 (2009)Google Scholar
  4. 4.
    Devasena, C.L., Sumathi, T., Hemulatha, M.: An Experiential Survey on Image Mining Tools, Techniques and Applications. Int. J. Computer Science and Engineering 3(3), 1155–1167 (2011)Google Scholar
  5. 5.
    Zhu, R., Yao, M., Ye, L.H., Xuan, J.Y.: Learning a Hierarchical Image Manifold for Web Image Classification. J. Zhejiang Univ.-Sci. C 13(10), 719–735 (2012)CrossRefGoogle Scholar
  6. 6.
    Biedeman, I.: Aspects and Extensions of a Theory of Human Image Understanding. In: Proc. of Computational Process in Human Vision: An Interdisciplinary Perspective, pp. 370–428 (1998)Google Scholar
  7. 7.
    Datta, R., Joshi, D., Li, J., Wang, J.Z.: Image Retrieval: Ideas, Influences, and Treads of the New Age. ACM Computing Surveys 40(2), 1–60 (2008)CrossRefGoogle Scholar
  8. 8.
    Seung, H.S., Lee, D.: The Manifold Ways of Perception. Science 290(5500), 2268–2269 (2000)CrossRefGoogle Scholar
  9. 9.
    Wang, S.J.: Bionic (Topological) Pattern Recognition- a New Model of Pattern Recognition Theory and Its Applications. Acta Electronica Sinica 30(10), 1417–1420 (2002)Google Scholar
  10. 10.
    Huo, X.M., Ni, X.L., Smith, A.K.: A Survey of Manifold-Based Learning Methods: Emerging Nonparametric Methodology (2007), http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.95.903
  11. 11.
    Kim, W., Chen, Y.C., Crawford, M.M., Tilton, J., Ghosh, J.: Multiresolution Manifold Learning for Classification of Hyperspectral Data (2009), http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.205.5783
  12. 12.
    Huh, S., Fienberg, S.: Discriminative Topic Modeling Based on Manifold Learning. In: ACM SIGKDD Conference on Knowledge Discover and Data MingGoogle Scholar
  13. 13.
    Farajtaba, M., Rabiee, H.R., Shaban, A., Soltani-Farani, A.: Efficient Iterative Semi-Supervised Classification on Manifold. In: IEEE International Conference on Data Mining, pp. 228–235 (2011)Google Scholar
  14. 14.
    Tax, D.M.J., Duin, R.P.W.: Using Two-Class Classifiers for Multiclass Classification. In: 16th International Conference on Pattern Recognition, vol. 2, pp. 124–127 (2002)Google Scholar
  15. 15.
    Aly, M.: Survey on Multiclass Classification Methods (2005), http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.175.107
  16. 16.
    Chen, W., Metz, C.E., Giger, M.L., Drukker, K.: A Novel Hybrid Linear/Nonlinear Classifier for Two-Class Classification: Theory, Algorithm, and Applications. IEEE Trans. on Medical Imaging 29(2), 428–441 (2010)CrossRefGoogle Scholar
  17. 17.
    Timotius, I.K., Setyawan, I., Febrianto, A.A.: Two-Class Classification with Various Characteristic Based on Kernel Principal Component Analysis and Support Vector Machines. Maraka J. Technology Series 15(1), 96–100 (2011)Google Scholar
  18. 18.
    Wen, W.H., Liu, G.Y., Xiong, X.: Feature Selection Model and Generalization Performance of Two-Class Emotion Recognition Systems Based on Physiological Signal. Computer Science 38(5), 220–223 (2011) (in Chinese)Google Scholar
  19. 19.
    Pan, W.G., Bao, H., He, N.: Novel Binary Classification Method for Traditional Chinese Paintings and Caligraphy Images. Computer Science 39(3), 257–260 (2012) (in Chinese)Google Scholar
  20. 20.
    Xu, Z.J., Yang, J., Wang, M.: A New Nonlinear Dimensionality Reduction for Color Image. J. Shanghai Jiaotong University 38(12), 2063–2072 (2004) (in Chinese)Google Scholar
  21. 21.
    Lu, D., Weng, Q.: A Survey of Image Classification Methods and Techniques for Improving Classification Performance. International J. Remote Sensor 28(5), 823–870 (2007)CrossRefGoogle Scholar
  22. 22.
    Jun, G., Ghosh, J.: Nearest-Manifold Classification with Gaussian Processes. In: 20th International Conference on Pattern Recognition, pp. 914–917 (2010)Google Scholar
  23. 23.
    Zhu, R., Yao, M.: Image Feature Optimization Based on Nonlinear Dimensionality Reduction. J. Zhejiang University Science A 10(12), 1720–1737 (2009)CrossRefzbMATHGoogle Scholar
  24. 24.
    Chang, H., Yeung, D.Y.: Robust Locally Linear Embedding. Pattern Recognition 39(6), 1053–1065 (2006)CrossRefzbMATHGoogle Scholar
  25. 25.
    Yao, L.Q., Tao, Q.: One Kind of Manifold Learning Method for Classification. PR & AI 18(5), 542–545 (2005) (in Chinese) Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  • Rong Zhu
    • 1
  • Jianhua Yang
    • 2
  • Yonggang Li
    • 1
  • Jie Xu
    • 1
  1. 1.School of Information EngineeringJiaxing UniversityJiaxingChina
  2. 2.School of Computer Science and TechnologyZhejiang UniversityHangzhouChina

Personalised recommendations